CPD Results
The following document contains the results of PMD's CPD 5.3.2.
Duplications
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ecb/ECBPlusEvaluator.java | pikes-resources | 25 |
eu/fbk/dkm/pikes/resources/ecb/ECBPlusEvaluatorLemma.java | pikes-resources | 25 |
private static final Logger LOGGER = LoggerFactory.getLogger(ECBPlusEvaluator.class); private static final Pattern tokenPattern = Pattern.compile(".*/([0-9]+)_([0-9]+ecb[a-z]*)\\.xml#char=([0-9]+).*"); private static final Pattern fileNamePattern = Pattern.compile("[0-9]+/([0-9]+)_([0-9a-zA-Z]+)"); // private static final Boolean removeAloneClusters = false; // private static final Pattern chainPattern = Pattern.compile("CHAIN=\"([0-9]+)\""); private static Integer FOLDER = null; public static void printToken(Appendable writer, Term token, int i, String last) throws IOException { writer.append(String.format("%d", i)).append("\t"); writer.append(token.getForm()).append("\t"); writer.append("_").append("\t"); writer.append(token.getForm()).append("\t"); writer.append("_").append("\t"); writer.append(token.getMorphofeat()).append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append(last); writer.append("\n"); } public static void main(String[] args) { try { final CommandLine cmd = CommandLine .parser() .withName("./ecb-evaluator") .withHeader("Evaluator event extractor") .withOption("n", "input-naf", "Input NAF folder", "FOLDER", CommandLine.Type.DIRECTORY_EXISTING, true, false, true) .withOption("i", "input-csv", "Input CSV file", "FILE", CommandLine.Type.FILE_EXISTING, true, false, true) .withOption("g", "output-gold", "Output gold file", "FILE", CommandLine.Type.FILE, true, false, true) .withOption("b", "output-baseline", "Output baseline file", "FILE", CommandLine.Type.FILE, true, false, true) .withOption("o", "output", "Output file", "FILE", CommandLine.Type.FILE, true, false, true) .withOption("l", "input-lemmas", "Lemmas CSV file", "FILE", CommandLine.Type.FILE_EXISTING, true, false, false) .withOption("a", "input-all-lemmas", "Lemmas CSV file", "FILE", CommandLine.Type.FILE_EXISTING, true, false, false) // .withOption("r", "remove-alone", "Remove alone clusters") .withOption("c", "check-gold", "Use only events annotated in gold standard") .withOption("s", "add-single", "Add single clusters") .withLogger(LoggerFactory.getLogger("eu.fbk")).parse(args); File inputCsv = cmd.getOptionValue("input-csv", File.class); File inputNaf = cmd.getOptionValue("input-naf", File.class); File inputLemmas = cmd.getOptionValue("input-lemmas", File.class); File inputAllLemmas = cmd.getOptionValue("input-all-lemmas", File.class); File outputGold = cmd.getOptionValue("output-gold", File.class); File outputBaseline = cmd.getOptionValue("output-baseline", File.class); File output = cmd.getOptionValue("output", File.class); // Boolean removeAloneClusters = cmd.hasOption("remove-alone"); Boolean checkGold = cmd.hasOption("check-gold"); Boolean addSingleClusters = cmd.hasOption("add-single"); Reader in; Iterable<CSVRecord> records; HashMap<String, Integer> lemmas = null; HashMap<String, Integer> allLemmas = null; int lemmaIndex = 0; if (inputLemmas != null) { lemmas = new HashMap<>(); in = new FileReader(inputLemmas); records = CSVFormat.EXCEL.withHeader().parse(in); for (CSVRecord record : records) { String lemma = record.get(1); lemma = lemma.replaceAll("\"", "").trim(); if (lemma.length() > 0) { lemmas.put(lemma, ++lemmaIndex); } } } lemmaIndex = 0; if (inputAllLemmas != null) { allLemmas = new HashMap<>(); in = new FileReader(inputAllLemmas); records = CSVFormat.EXCEL.withHeader().parse(in); for (CSVRecord record : records) { String lemma = record.get(1); lemma = lemma.replaceAll("\"", "").trim(); if (lemma.length() > 0) { allLemmas.put(lemma, ++lemmaIndex); } } } if (lemmas != null) { LOGGER.info("Lemmas: {}", lemmas.size()); } if (allLemmas != null) { LOGGER.info("All-lemmas: {}", allLemmas.size()); } BufferedWriter goldWriter = new BufferedWriter(new FileWriter(outputGold)); BufferedWriter baselineWriter = new BufferedWriter(new FileWriter(outputBaseline)); BufferedWriter writer = new BufferedWriter(new FileWriter(output)); HashMultimap<String, String> goldTmpClusters = HashMultimap.create(); HashMap<String, String> goldClusters = new HashMap<>(); Set<String> okEvents = new HashSet<>(); Map<String, String> theBaseline = new HashMap<>(); for (final File file : Files.fileTreeTraverser().preOrderTraversal(inputNaf)) { if (!file.isFile()) { continue; } if (file.getName().startsWith(".")) { continue; } String path = file.getParentFile().toString(); String relativeFilePath = file.getAbsolutePath() .substring(inputNaf.getAbsolutePath().length()); Matcher matcher = fileNamePattern.matcher(relativeFilePath); Integer folder = null; String fileNum = null; if (matcher.find()) { folder = Integer.parseInt(matcher.group(1)); fileNum = matcher.group(2); } else { LOGGER.error("Error in file name: {}", relativeFilePath); System.exit(1); } if (FOLDER != null && !folder.equals(FOLDER)) { continue; } LOGGER.debug(file.getAbsolutePath()); KAFDocument document = KAFDocument.createFromFile(file); for (Coref coref : document.getCorefs()) { if (coref.getType() == null) { continue; } if (!coref.getType().equals("event-gold")) { continue; } Integer cluster = Integer.parseInt(coref.getCluster()); String idCluster = String.valueOf(1000 * folder + cluster); for (Span<Term> termSpan : coref.getSpans()) { Term term = termSpan.getTargets().get(0); String lemma = term.getLemma(); boolean add = false; if (allLemmas != null && allLemmas.containsKey(lemma)) { add = true; } if (lemmas == null || lemmas.containsKey(lemma)) { add = true; } if (add) { String text = folder + "_" + fileNum + "_" + term.getOffset(); goldTmpClusters.put(idCluster, text); goldClusters.put(text, idCluster); okEvents.add(text); } } } goldWriter.append(String.format("#begin document %d_%s", folder, fileNum)).append("\n"); baselineWriter.append(String.format("#begin document %d_%s", folder, fileNum)).append("\n"); Integer numSentences = document.getNumSentences(); for (int i = 1; i <= numSentences; i++) { boolean useThis = false; StringBuilder goldBuilder = new StringBuilder(); StringBuilder baselineBuilder = new StringBuilder(); List<Term> sentenceTerms = document.getSentenceTerms(i); int n = 0; for (Term token : sentenceTerms) { String id = String.format("%d_%s_%d", folder, fileNum, token.getOffset()); String last; n++; last = "_"; if (goldClusters.containsKey(id)) { last = String.format("(%s)", goldClusters.get(id)); useThis = true; } printToken(goldBuilder, token, n, last); last = "_"; String lemma = token.getLemma(); if (lemmas != null) { if (goldClusters.containsKey(id) && lemmas.containsKey(lemma)) { last = String.format("(%d)", lemmas.get(lemma)); } } if (allLemmas != null) { if (goldClusters.containsKey(id) && allLemmas.containsKey(lemma)) { last = String.format("(%d)", allLemmas.get(lemma)); } } if (!last.equals("_")) { theBaseline.put(id, last); } printToken(baselineBuilder, token, n, last); } goldBuilder.append("\n"); baselineBuilder.append("\n"); if (useThis) { goldWriter.append(goldBuilder.toString()); baselineWriter.append(baselineBuilder.toString()); } } // break; } goldWriter.close(); baselineWriter.close(); // Set<Set> goldClusters = new HashSet<>(); // for (String key : goldTmpClusters.keySet()) { // Set<String> cluster = goldTmpClusters.get(key); // if (cluster.size() > 1 || !removeAloneClusters) { // goldClusters.add(cluster); // } // } // LOGGER.info("Gold clusters: {}", goldClusters.size()); in = new FileReader(inputCsv); records = CSVFormat.EXCEL.withHeader().parse(in); // Size must be always 4! int clusterID = 0; HashMap<String, Integer> clusterIndexes = new HashMap<>(); HashMultimap<Integer, String> theClusters = HashMultimap.create(); for (CSVRecord record : records) { Matcher matcher; String id1 = null; String id2 = null; matcher = tokenPattern.matcher(record.get(1)); if (matcher.find()) { id1 = matcher.group(1) + "_" + matcher.group(2) + "_" + matcher.group(3); } matcher = tokenPattern.matcher(record.get(3)); if (matcher.find()) { id2 = matcher.group(1) + "_" + matcher.group(2) + "_" + matcher.group(3); } // System.out.println(id1); // System.out.println(id2); Integer index1 = clusterIndexes.get(id1); Integer index2 = clusterIndexes.get(id2); // System.out.println(index1); // System.out.println(index2); if (index1 == null && index2 == null) { clusterID++; if (!checkGold || okEvents.contains(id2)) { if (id2 != null) { theClusters.put(clusterID, id2); clusterIndexes.put(id2, clusterID); } } if (!checkGold || okEvents.contains(id1)) { if (id1 != null) { theClusters.put(clusterID, id1); clusterIndexes.put(id1, clusterID); } } } if (index1 == null && index2 != null) { if (!checkGold || okEvents.contains(id1)) { if (id1 != null) { theClusters.put(index2, id1); clusterIndexes.put(id1, index2); } } } if (index2 == null && index1 != null) { if (!checkGold || okEvents.contains(id2)) { if (id2 != null) { theClusters.put(index1, id2); clusterIndexes.put(id2, index1); } } } if (index2 != null && index1 != null) { if (!index1.equals(index2)) { if (id2 != null) { clusterIndexes.put(id2, index1); theClusters.putAll(index1, theClusters.get(index2)); theClusters.removeAll(index2); } } } } // System.out.println(theClusters); // System.out.println(theBaseline); int otherClusterID = 100000; for (final File file : Files.fileTreeTraverser().preOrderTraversal(inputNaf)) { if (!file.isFile()) { continue; } if (file.getName().startsWith(".")) { continue; } // String path = file.getParentFile().toString(); String relativeFilePath = file.getAbsolutePath() .substring(inputNaf.getAbsolutePath().length()); Matcher matcher = fileNamePattern.matcher(relativeFilePath); Integer folder = null; String fileNum = null; if (matcher.find()) { folder = Integer.parseInt(matcher.group(1)); fileNum = matcher.group(2); } else { LOGGER.error("Error in file name: {}", relativeFilePath); System.exit(1); } // Integer folder = Integer.parseInt(path.substring(path.lastIndexOf("/")).substring(1)); // Integer fileNum = Integer.parseInt(file.getName().substring(0, file.getName().length() - 4)); LOGGER.debug(file.getAbsolutePath()); KAFDocument document = KAFDocument.createFromFile(file); if (FOLDER != null && !folder.equals(FOLDER)) { continue; } writer.append(String.format("#begin document %d_%s", folder, fileNum)).append("\n"); Integer numSentences = document.getNumSentences(); for (int i = 1; i <= numSentences; i++) { boolean useThis = false; StringBuilder outBuilder = new StringBuilder(); List<Term> sentenceTerms = document.getSentenceTerms(i); int n = 0; for (Term token : sentenceTerms) { String id = String.format("%d_%s_%d", folder, fileNum, token.getOffset()); if (okEvents.contains(id)) { useThis = true; } String last = theBaseline.getOrDefault(id, "_"); if (clusterIndexes.containsKey(id)) { last = String.format("(%d)", clusterIndexes.get(id) + 1000000); } if (last.equals("_")) { if (okEvents.contains(id) && addSingleClusters) { last = String.format("(%d)", ++otherClusterID); } } printToken(outBuilder, token, ++n, last); } outBuilder.append("\n"); if (useThis) { writer.append(outBuilder.toString()); } } } writer.close(); } catch (Exception e) { CommandLine.fail(e); } } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/raid/mdfsa/APIManager.java | pikes-raid | 61 |
eu/fbk/dkm/pikes/raid/mdfsa/APIManager.java | pikes-raid | 294 |
HashMap<String, Integer> sentenceTokens = new HashMap<String, Integer>(); HashMap<String, Integer> stemmedTokens = new HashMap<String, Integer>(); FuzzyMembership eT = new FuzzyMembership(1.0, 1.0, -1.0, -1.0); String[] tokens = text.split(" "); for(int i = 0; i < tokens.length; i++) { if(tokens[i].compareTo("") != 0) { String lexToken = this.wnlex.getWordLexicalizationByType(tokens[i].toLowerCase(), "MIX"); //System.out.print(" " + lexToken + " "); if(lexToken != null) { tokens[i] = lexToken; } sentenceTokens.put(tokens[i], new Integer(i)); porterStemmer stemmer = new porterStemmer(); stemmer.setCurrent(tokens[i].toLowerCase()); String stemLink = tokens[i].toLowerCase(); if(stemmer.stem()) { stemLink = stemmer.getCurrent(); //System.out.print(stemLink + " "); } stemmedTokens.put(stemLink, new Integer(i)); } } Iterator<String> it = this.labels.keySet().iterator(); while(it.hasNext()) { String currentConcept = it.next(); String[] cts = currentConcept.split("_"); int higherIdx = 0; int lowerIdx = tokens.length; int foundCT = 0; int flagNegation = 1; for(String ct: cts) { /* if(ct.compareTo("regret") == 0) { int A = 1; } */ Integer tempIdx = sentenceTokens.get(ct); if(tempIdx == null) { tempIdx = stemmedTokens.get(ct); } if(tempIdx != null) { if(tempIdx < lowerIdx) lowerIdx = tempIdx; if(tempIdx > higherIdx) higherIdx = tempIdx; foundCT++; } } Integer notToken = sentenceTokens.get("not"); if(notToken == null) { //notToken = sentenceTokens.get("no"); } if(notToken != null && ( (notToken >= (lowerIdx - 2)) || ( (notToken > lowerIdx) && (notToken < higherIdx) ) ) ) { flagNegation = -1; } /* if(notToken != null && notToken == lowerIdx - 1) { flagNegation = -1; } */ if(higherIdx >= 0 && foundCT == cts.length && (higherIdx - lowerIdx) < (cts.length + 2)) { Long feature = this.labels.get(currentConcept); //Long feature = labels.get(stemLink); /* double ratioFactor = 1.0 / (sentenceTokens.size() - lowerIdx); if(ratioFactor == Double.NaN || ratioFactor == 0.0) { ratioFactor = 1.0; } */ double ratioFactor = 1.0; FuzzyMembership fm = this.polarities.get(feature); double a = 0.0; try { a = fm.getA() * ratioFactor * flagNegation; } catch (Exception e) { System.out.println("Error on getting fuzzy shape: " + currentConcept); //System.exit(0); return -2.0; } double b = fm.getB() * ratioFactor * flagNegation; double c = fm.getC() * ratioFactor * flagNegation; double d = fm.getD() * ratioFactor * flagNegation; if(flagNegation == -1) { double t = a; a = d; d = t; t = b; b = c; c = t; } double eA = eT.getA(); double eB = eT.getB(); double eC = eT.getC(); double eD = eT.getD(); if(a < eA) eA = a; if(b < eB) eB = b; if(c > eC) eC = c; if(d > eD) eD = d; eT.setA(eA); eT.setB(eB); eT.setC(eC); eT.setD(eD); fuzzyShapeFound += 1.0; tempPolarity += fm.getCentroidXAxis(); } } if(eT.getA() != 1.0) { //textPolarity = eT.getCentroid(); textPolarity = eT.getCentroidXAxis(); if(Double.isNaN(textPolarity)) { return -2.0; } //textPolarity = tempPolarity / fuzzyShapeFound; } return textPolarity; } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/raid/pipeline/PipelineTrainer.java | pikes-raid | 152 |
eu/fbk/dkm/pikes/raid/sbrs/SBRSTrainer.java | pikes-raid | 76 |
} private void addArguments(final KAFDocument document, final int sentence, final Opinion[] opinions) { // Index holder and target spans by expression head, keeping track of all exp. heads final Set<Term> expressionHeads = Sets.newHashSet(); final Multimap<Term, Span<Term>> holderSpans = HashMultimap.create(); final Multimap<Term, Span<Term>> targetSpans = HashMultimap.create(); for (final Opinion opinion : opinions) { final Set<Term> heads = Opinions.heads(document, NAFUtils.normalizeSpan(document, opinion.getExpressionSpan()), Component.EXPRESSION); if (!heads.isEmpty()) { final Term head = Ordering.from(Term.OFFSET_COMPARATOR).max(heads); expressionHeads.add(head); final Span<Term> holderSpan = opinion.getHolderSpan(); final Span<Term> targetSpan = opinion.getTargetSpan(); if (holderSpan != null) { holderSpans.putAll(head, NAFUtils.splitSpan(document, holderSpan, // Opinions.heads(document, holderSpan, Component.HOLDER))); } if (targetSpan != null) { targetSpans.putAll(head, NAFUtils.splitSpan(document, targetSpan, // Opinions.heads(document, targetSpan, Component.TARGET))); } } } // Add training samples for holder and target extraction, separately (if enabled) for (final Term expressionHead : expressionHeads) { if (components().contains(Component.HOLDER)) { addArguments(document, sentence, expressionHead, holderSpans.get(expressionHead), this.holderLinkTrainer, this.holderSpanTrainer); } if (components().contains(Component.TARGET)) { addArguments(document, sentence, expressionHead, targetSpans.get(expressionHead), this.targetLinkTrainer, this.targetSpanTrainer); } } } private void addArguments(final KAFDocument document, final int sentence, final Term expressionHead, final Iterable<Span<Term>> argSpans, final LinkLabeller.Trainer linkTrainer, final SpanLabeller.Trainer spanTrainer) { // Extract heads and spans of the arguments (only where defined) final List<Term> heads = Lists.newArrayList(); final List<Span<Term>> spans = Lists.newArrayList(); for (final Span<Term> span : argSpans) { final Term head = NAFUtils.extractHead(document, span); if (head != null) { heads.add(head); spans.add(span); } } // Add a sample for node labelling linkTrainer.add(document, expressionHead, heads); // Add samples for span labelling (one for each argument) for (int i = 0; i < heads.size(); ++i) { final List<Term> excludedTerms = Lists.newArrayList(heads); excludedTerms.remove(heads.get(i)); spanTrainer.add(document, heads.get(i), excludedTerms, spans.get(i)); } } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/trec/CR.java | pikes-resources | 45 |
eu/fbk/dkm/pikes/resources/trec/LATIMES.java | pikes-resources | 49 |
.withHeader("Extract CR documents from TREC dataset and save them in NAF format") .withOption("i", "input", "Input folder", "FOLDER", CommandLine.Type.DIRECTORY_EXISTING, true, false, true) .withOption("o", "output", "Output folder", "FOLDER", CommandLine.Type.DIRECTORY, true, false, true) .withOption("u", "url-template", "URL template (with %d for the ID)", "URL", CommandLine.Type.STRING, true, false, false) .withLogger(LoggerFactory.getLogger("eu.fbk")) // .parse(args); File inputDir = cmd.getOptionValue("input", File.class); String urlTemplate = DEFAULT_URL; if (cmd.hasOption("url-template")) { urlTemplate = cmd.getOptionValue("url-template", String.class); } File outputDir = cmd.getOptionValue("output", File.class); if (!outputDir.exists()) { outputDir.mkdirs(); } for (final File file : Files.fileTreeTraverser().preOrderTraversal(inputDir)) { if (!file.isFile()) { continue; } if (file.getName().startsWith(".")) { continue; } String outputTemplate = outputDir.getAbsolutePath() + File.separator + file.getName(); File newFolder = new File(outputTemplate); newFolder.mkdirs(); outputTemplate += File.separator + "NAF"; saveFile(file, outputTemplate, urlTemplate); } } catch (Exception e) { CommandLine.fail(e); } } private static void saveFile(File inputFile, String outputFilePattern, String urlTemplate) throws IOException, SAXException, ParserConfigurationException { LOGGER.info("Input file: {}", inputFile); StringBuffer stringBuffer = new StringBuffer(); stringBuffer.append("<ROOT>\n"); stringBuffer.append(Files.toString(inputFile, Charsets.UTF_8)); stringBuffer.append("\n</ROOT>\n"); InputStream is = new ByteArrayInputStream(stringBuffer.toString().getBytes()); DocumentBuilderFactory dbFactory = DocumentBuilderFactory.newInstance(); DocumentBuilder dBuilder = dbFactory.newDocumentBuilder(); Document doc = dBuilder.parse(is); doc.getDocumentElement().normalize(); int i = 0; for (Element element : JOOX.$(doc).find("DOC")) { Element docnoElement = JOOX.$(element).find("DOCNO").get(0); Element dateElement = JOOX.$(element).find("DATE").get(0); Element headlineElement = JOOX.$(element).find("TTL").get(0); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ecb/ECBPlusEvaluator.java | pikes-resources | 27 |
eu/fbk/dkm/pikes/resources/ecb/ECBPlusEvaluatorLemma.java | pikes-resources | 27 |
eu/fbk/dkm/pikes/resources/ecb/ECBevaluator.java | pikes-resources | 29 |
private static final Pattern fileNamePattern = Pattern.compile("[0-9]+/([0-9]+)_([0-9a-zA-Z]+)"); // private static final Boolean removeAloneClusters = false; // private static final Pattern chainPattern = Pattern.compile("CHAIN=\"([0-9]+)\""); private static Integer FOLDER = null; public static void printToken(Appendable writer, Term token, int i, String last) throws IOException { writer.append(String.format("%d", i)).append("\t"); writer.append(token.getForm()).append("\t"); writer.append("_").append("\t"); writer.append(token.getForm()).append("\t"); writer.append("_").append("\t"); writer.append(token.getMorphofeat()).append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append(last); writer.append("\n"); } public static void main(String[] args) { try { final CommandLine cmd = CommandLine .parser() .withName("./ecb-evaluator") .withHeader("Evaluator event extractor") .withOption("n", "input-naf", "Input NAF folder", "FOLDER", CommandLine.Type.DIRECTORY_EXISTING, true, false, true) .withOption("i", "input-csv", "Input CSV file", "FILE", CommandLine.Type.FILE_EXISTING, true, false, true) .withOption("g", "output-gold", "Output gold file", "FILE", CommandLine.Type.FILE, true, false, true) .withOption("b", "output-baseline", "Output baseline file", "FILE", CommandLine.Type.FILE, true, false, true) .withOption("o", "output", "Output file", "FILE", CommandLine.Type.FILE, true, false, true) .withOption("l", "input-lemmas", "Lemmas CSV file", "FILE", CommandLine.Type.FILE_EXISTING, true, false, false) .withOption("a", "input-all-lemmas", "Lemmas CSV file", "FILE", |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/tintopclient/TintopOrchestrator.java | pikes-tintop-client | 169 |
eu/fbk/dkm/pikes/tintop/FolderOrchestrator.java | pikes-tintop | 151 |
} synchronized public String getNextFile(TintopSession session) { fIter: while (fileCache != null || session.getFileIterator().hasNext()) { File file; if (fileCache != null) { file = new File(fileCache); fileCache = null; } else { file = session.getFileIterator().next(); } File outputFile = getOutputFile(file, session); // todo: use parameters outputFile = new File(outputFile.getAbsolutePath() + ".gz"); logger.debug("Output file: " + outputFile); if (outputFile.exists()) { logger.debug("Skipping file (it exists): " + file); continue fIter; } for (String p : session.getSkipPatterns()) { if (file.toString().contains(p)) { logger.debug("Skipping file (skip pattern): " + file); continue fIter; } } if (maxSize > 0 && file.length() > maxSize) { logger.debug("Skipping file (too big, " + file.length() + "): " + file); skipped++; continue fIter; } // File is empty if (file.length() < 1000) { try { KAFDocument document = KAFDocument.createFromFile(file); if (document.getRawText() == null || document.getRawText().trim().length() == 0) { logger.info("File is empty: " + file); logger.info("Writing empty file " + outputFile); Files.createParentDirs(outputFile); try (Writer w = IO.utf8Writer(IO.buffer(IO.write(outputFile.getAbsolutePath())))) { w.write(document.toString()); } continue fIter; } } catch (IOException e) { e.printStackTrace(); skipped++; continue fIter; } catch (JDOMException e) { e.printStackTrace(); skipped++; continue fIter; } } return file.getAbsolutePath(); } return null; } public void run(TintopSession session) { |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ontonotes/senses/Sense.java | pikes-resources | 19 |
eu/fbk/dkm/pikes/resources/util/onsenses/Sense.java | pikes-resources | 24 |
@XmlAccessorType(XmlAccessType.FIELD) @XmlType(name = "", propOrder = { "commentary", "examples", "mappings", "sensemeta" }) @XmlRootElement(name = "sense") public class Sense { @XmlAttribute(name = "n", required = true) @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String n; @XmlAttribute(name = "type") @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String type; @XmlAttribute(name = "name", required = true) @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String name; @XmlAttribute(name = "group", required = true) @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String group; protected String commentary; @XmlElement(required = true) protected String examples; @XmlElement(required = true) protected Mappings mappings; @XmlElement(name = "SENSE_META", required = true) protected SENSEMETA sensemeta; /** * Recupera il valore della proprietà n. * * @return * possible object is * {@link String } * */ public String getN() { return n; } /** * Imposta il valore della proprietà n. * * @param value * allowed object is * {@link String } * */ public void setN(String value) { this.n = value; } /** * Recupera il valore della proprietà type. * * @return * possible object is * {@link String } * */ public String getType() { return type; } /** * Imposta il valore della proprietà type. * * @param value * allowed object is * {@link String } * */ public void setType(String value) { this.type = value; } /** * Recupera il valore della proprietà name. * * @return * possible object is * {@link String } * */ public String getName() { return name; } /** * Imposta il valore della proprietà name. * * @param value * allowed object is * {@link String } * */ public void setName(String value) { this.name = value; } /** * Recupera il valore della proprietà group. * * @return * possible object is * {@link String } * */ public String getGroup() { return group; } /** * Imposta il valore della proprietà group. * * @param value * allowed object is * {@link String } * */ public void setGroup(String value) { this.group = value; } /** * Recupera il valore della proprietà commentary. * * @return * possible object is * {@link String } * */ public String getCommentary() { return commentary; } /** * Imposta il valore della proprietà commentary. * * @param value * allowed object is * {@link String } * */ public void setCommentary(String value) { this.commentary = value; } /** * Recupera il valore della proprietà examples. * * @return * possible object is * {@link String } * */ public String getExamples() { return examples; } /** * Imposta il valore della proprietà examples. * * @param value * allowed object is * {@link String } * */ public void setExamples(String value) { this.examples = value; } /** * Recupera il valore della proprietà mappings. * * @return * possible object is * {@link Mappings } * */ public Mappings getMappings() { return mappings; } /** * Imposta il valore della proprietà mappings. * * @param value * allowed object is * {@link Mappings } * */ public void setMappings(Mappings value) { this.mappings = value; } /** * Recupera il valore della proprietà sensemeta. * * @return * possible object is * {@link SENSEMETA } * */ public SENSEMETA getSENSEMETA() { return sensemeta; } /** * Imposta il valore della proprietà sensemeta. * * @param value * allowed object is * {@link SENSEMETA } * */ public void setSENSEMETA(SENSEMETA value) { this.sensemeta = value; } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/rdf/Renderer.java | pikes-rdf | 480 |
eu/fbk/dkm/pikes/rdf/naf/NAFRenderer.java | pikes-rdf | 421 |
final URI mentionURI = (URI) stmt.getObject(); final String name = mentionURI.getLocalName(); if (name.indexOf(';') < 0) { final int index = name.indexOf(','); final int start = Integer.parseInt(name.substring(5, index)); final int end = Integer.parseInt(name.substring(index + 1)); final int s = Arrays.binarySearch(offsets, start); if (s >= 0) { int e = s; while (e < offsets.length && offsets[e] < end) { ++e; } markables.add(new Markable(ImmutableList.copyOf(terms.subList(s, e)), color)); } } } } return markables; } private static String select(final Map<Object, String> map, final Iterable<? extends Value> keys, final String defaultColor) { String color = null; for (final Value key : keys) { if (key instanceof URI) { final String mappedColor = map.get(key); if (mappedColor != null) { if (color == null) { color = mappedColor; } else { break; } } } } return color != null ? color : defaultColor; } private static String escape(final String string) { return HtmlEscapers.htmlEscaper().escape(string); } @Nullable private static String shorten(@Nullable final URI uri) { if (uri == null) { return null; } final String prefix = Namespaces.DEFAULT.prefixFor(uri.getNamespace()); if (prefix != null) { return prefix + ':' + uri.getLocalName(); } return "<../" + uri.getLocalName() + ">"; } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ontonotes/frames/Roleset.java | pikes-resources | 22 |
eu/fbk/dkm/pikes/resources/util/propbank/Roleset.java | pikes-resources | 20 |
@XmlAccessorType(XmlAccessType.FIELD) @XmlType(name = "", propOrder = { "noteOrRolesOrExample" }) @XmlRootElement(name = "roleset") public class Roleset { @XmlAttribute(name = "id", required = true) @XmlJavaTypeAdapter(CollapsedStringAdapter.class) @XmlID protected String id; @XmlAttribute(name = "name") @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String name; @XmlAttribute(name = "vncls") @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String vncls; @XmlAttribute(name = "framnet") @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String framnet; @XmlAttribute(name = "source") @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String source; @XmlElements({ @XmlElement(name = "note", type = Note.class), @XmlElement(name = "roles", type = Roles.class), @XmlElement(name = "example", type = Example.class) }) protected List<Object> noteOrRolesOrExample; /** * Recupera il valore della proprietà id. * * @return * possible object is * {@link String } * */ public String getId() { return id; } /** * Imposta il valore della proprietà id. * * @param value * allowed object is * {@link String } * */ public void setId(String value) { this.id = value; } /** * Recupera il valore della proprietà name. * * @return * possible object is * {@link String } * */ public String getName() { return name; } /** * Imposta il valore della proprietà name. * * @param value * allowed object is * {@link String } * */ public void setName(String value) { this.name = value; } /** * Recupera il valore della proprietà vncls. * * @return * possible object is * {@link String } * */ public String getVncls() { return vncls; } /** * Imposta il valore della proprietà vncls. * * @param value * allowed object is * {@link String } * */ public void setVncls(String value) { this.vncls = value; } /** * Recupera il valore della proprietà framnet. * * @return * possible object is * {@link String } * */ public String getFramnet() { return framnet; } /** * Imposta il valore della proprietà framnet. * * @param value * allowed object is * {@link String } * */ public void setFramnet(String value) { this.framnet = value; } /** * Recupera il valore della proprietà source. * * @return * possible object is * {@link String } * */ public String getSource() { return source; } /** * Imposta il valore della proprietà source. * * @param value * allowed object is * {@link String } * */ public void setSource(String value) { this.source = value; } /** * Gets the value of the noteOrRolesOrExample property. * * <p> * This accessor method returns a reference to the live list, * not a snapshot. Therefore any modification you make to the * returned list will be present inside the JAXB object. * This is why there is not a <CODE>set</CODE> method for the noteOrRolesOrExample property. * * <p> * For example, to add a new item, do as follows: * <pre> * getNoteOrRolesOrExample().add(newItem); * </pre> * * * <p> * Objects of the following type(s) are allowed in the list * {@link Note } * {@link Roles } * {@link Example } * * */ public List<Object> getNoteOrRolesOrExample() { if (noteOrRolesOrExample == null) { noteOrRolesOrExample = new ArrayList<Object>(); } return this.noteOrRolesOrExample; } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/trec/FBIS.java | pikes-resources | 50 |
eu/fbk/dkm/pikes/resources/trec/FR94.java | pikes-resources | 47 |
.withHeader("Extract FBIS documents from TREC dataset and save them in NAF format") .withOption("i", "input", "Input folder", "FOLDER", CommandLine.Type.DIRECTORY_EXISTING, true, false, true) .withOption("o", "output", "Output folder", "FOLDER", CommandLine.Type.DIRECTORY, true, false, true) .withOption("u", "url-template", "URL template (with %d for the ID)", "URL", CommandLine.Type.STRING, true, false, false) .withLogger(LoggerFactory.getLogger("eu.fbk")) // .parse(args); File inputDir = cmd.getOptionValue("input", File.class); String urlTemplate = DEFAULT_URL; if (cmd.hasOption("url-template")) { urlTemplate = cmd.getOptionValue("url-template", String.class); } File outputDir = cmd.getOptionValue("output", File.class); if (!outputDir.exists()) { outputDir.mkdirs(); } for (final File file : Files.fileTreeTraverser().preOrderTraversal(inputDir)) { if (!file.isFile()) { continue; } if (file.getName().startsWith(".")) { continue; } String outputTemplate = outputDir.getAbsolutePath() + File.separator + file.getName(); File newFolder = new File(outputTemplate); newFolder.mkdirs(); outputTemplate += File.separator + "NAF"; saveFile(file, outputTemplate, urlTemplate); } } catch (Exception e) { CommandLine.fail(e); } } private static void saveFile(File inputFile, String outputFilePattern, String urlTemplate) throws IOException, SAXException, ParserConfigurationException { LOGGER.info("Input file: {}", inputFile); StringBuffer stringBuffer = new StringBuffer(); stringBuffer.append("<?xml version=\"1.0\"?>\n" + "<!DOCTYPE tutorials [\n"); stringBuffer.append("<!ENTITY amp \" \">\n"); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/trec/CR.java | pikes-resources | 45 |
eu/fbk/dkm/pikes/resources/trec/FBIS.java | pikes-resources | 50 |
eu/fbk/dkm/pikes/resources/trec/FR94.java | pikes-resources | 47 |
eu/fbk/dkm/pikes/resources/trec/LATIMES.java | pikes-resources | 49 |
.withHeader("Extract CR documents from TREC dataset and save them in NAF format") .withOption("i", "input", "Input folder", "FOLDER", CommandLine.Type.DIRECTORY_EXISTING, true, false, true) .withOption("o", "output", "Output folder", "FOLDER", CommandLine.Type.DIRECTORY, true, false, true) .withOption("u", "url-template", "URL template (with %d for the ID)", "URL", CommandLine.Type.STRING, true, false, false) .withLogger(LoggerFactory.getLogger("eu.fbk")) // .parse(args); File inputDir = cmd.getOptionValue("input", File.class); String urlTemplate = DEFAULT_URL; if (cmd.hasOption("url-template")) { urlTemplate = cmd.getOptionValue("url-template", String.class); } File outputDir = cmd.getOptionValue("output", File.class); if (!outputDir.exists()) { outputDir.mkdirs(); } for (final File file : Files.fileTreeTraverser().preOrderTraversal(inputDir)) { if (!file.isFile()) { continue; } if (file.getName().startsWith(".")) { continue; } String outputTemplate = outputDir.getAbsolutePath() + File.separator + file.getName(); File newFolder = new File(outputTemplate); newFolder.mkdirs(); outputTemplate += File.separator + "NAF"; saveFile(file, outputTemplate, urlTemplate); } } catch (Exception e) { CommandLine.fail(e); } } private static void saveFile(File inputFile, String outputFilePattern, String urlTemplate) throws IOException, SAXException, ParserConfigurationException { LOGGER.info("Input file: {}", inputFile); StringBuffer stringBuffer = new StringBuffer(); stringBuffer.append("<ROOT>\n"); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/rdf/Renderer.java | pikes-rdf | 287 |
eu/fbk/dkm/pikes/rdf/naf/NAFRenderer.java | pikes-rdf | 220 |
private void renderPropertiesHelper(final Appendable out, final Model model, final Resource node, final boolean emitID, final Set<Resource> seen, final Set<URI> excludedProperties) throws IOException { // Open properties table out.append("<table class=\"properties table table-condensed\">\n<tbody>\n"); // Emit a raw for the node ID, if requested if (emitID) { out.append("<tr><td><a>ID</a>:</td><td>"); renderObject(out, node, model); out.append("</td></tr>\n"); } // Emit other properties for (final URI pred : this.valueComparator.sortedCopy(model.filter(node, null, null) .predicates())) { if (excludedProperties.contains(pred)) { continue; } out.append("<tr><td>"); renderObject(out, pred, model); out.append(":</td><td>"); final List<Resource> nested = Lists.newArrayList(); String separator = ""; for (final Value obj : this.valueComparator.sortedCopy(model.filter(node, pred, null) .objects())) { if (obj instanceof Literal || model.filter((Resource) obj, null, null).isEmpty()) { out.append(separator); renderObject(out, obj, model); separator = ", "; } else { nested.add((Resource) obj); } } out.append("".equals(separator) ? "" : "<br/>"); for (final Resource obj : nested) { out.append(separator); if (seen.add(obj)) { renderPropertiesHelper(out, model, obj, true, seen, excludedProperties); } else { renderObject(out, obj, model); } } out.append("</td></tr>\n"); } // Close properties table out.append("</tbody>\n</table>\n"); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ontonotes/frames/Inflection.java | pikes-resources | 19 |
eu/fbk/dkm/pikes/resources/util/propbank/Inflection.java | pikes-resources | 17 |
@XmlAccessorType(XmlAccessType.FIELD) @XmlType(name = "") @XmlRootElement(name = "inflection") public class Inflection { @XmlAttribute(name = "person") @XmlJavaTypeAdapter(CollapsedStringAdapter.class) protected String person; @XmlAttribute(name = "tense") @XmlJavaTypeAdapter(CollapsedStringAdapter.class) protected String tense; @XmlAttribute(name = "aspect") @XmlJavaTypeAdapter(CollapsedStringAdapter.class) protected String aspect; @XmlAttribute(name = "voice") @XmlJavaTypeAdapter(CollapsedStringAdapter.class) protected String voice; @XmlAttribute(name = "form") @XmlJavaTypeAdapter(CollapsedStringAdapter.class) protected String form; /** * Recupera il valore della proprietà person. * * @return * possible object is * {@link String } * */ public String getPerson() { if (person == null) { return "ns"; } else { return person; } } /** * Imposta il valore della proprietà person. * * @param value * allowed object is * {@link String } * */ public void setPerson(String value) { this.person = value; } /** * Recupera il valore della proprietà tense. * * @return * possible object is * {@link String } * */ public String getTense() { if (tense == null) { return "ns"; } else { return tense; } } /** * Imposta il valore della proprietà tense. * * @param value * allowed object is * {@link String } * */ public void setTense(String value) { this.tense = value; } /** * Recupera il valore della proprietà aspect. * * @return * possible object is * {@link String } * */ public String getAspect() { if (aspect == null) { return "ns"; } else { return aspect; } } /** * Imposta il valore della proprietà aspect. * * @param value * allowed object is * {@link String } * */ public void setAspect(String value) { this.aspect = value; } /** * Recupera il valore della proprietà voice. * * @return * possible object is * {@link String } * */ public String getVoice() { if (voice == null) { return "ns"; } else { return voice; } } /** * Imposta il valore della proprietà voice. * * @param value * allowed object is * {@link String } * */ public void setVoice(String value) { this.voice = value; } /** * Recupera il valore della proprietà form. * * @return * possible object is * {@link String } * */ public String getForm() { if (form == null) { return "ns"; } else { return form; } } /** * Imposta il valore della proprietà form. * * @param value * allowed object is * {@link String } * */ public void setForm(String value) { this.form = value; } } |
File | Project | Line |
---|---|---|
ixa/kaflib/Opinion.java | pikes-naflib | 78 |
ixa/kaflib/Opinion.java | pikes-naflib | 195 |
for (final ExternalRef externalRef : oh.getExternalRefs()) { this.externalReferences.add(new ExternalRef(externalRef)); } } public boolean hasType() { return this.type != null; } public String getType() { return this.type; } public void setType(final String type) { this.type = type; } public List<Term> getTerms() { return this.span.getTargets(); } public void addTerm(final Term term) { this.span.addTarget(term); } public void addTerm(final Term term, final boolean isHead) { this.span.addTarget(term, isHead); } public Span<Term> getSpan() { return this.span; } public void setSpan(final Span<Term> span) { this.span = span; } public ExternalRef getExternalRef(final String resource) { for (final ExternalRef ref : this.externalReferences) { if (ref.getResource().equalsIgnoreCase(resource)) { return ref; } } return null; } public List<ExternalRef> getExternalRefs() { return this.externalReferences; } public void addExternalRef(final ExternalRef externalRef) { this.externalReferences.add(externalRef); } public void addExternalRefs(final List<ExternalRef> externalRefs) { this.externalReferences.addAll(externalRefs); } @Override public boolean equals(final Object object) { if (object == this) { return true; } if (!(object instanceof OpinionTarget)) { return false; } final OpinionHolder other = (OpinionHolder) object; |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/rdf/Renderer.java | pikes-rdf | 554 |
eu/fbk/dkm/pikes/rdf/naf/NAFRenderer.java | pikes-rdf | 527 |
@Nullable private Map<Object, String> colorMap; @Nullable private Map<Object, String> styleMap; @Nullable private Mustache template; private final Map<String, Object> templateParameters; Builder() { this.templateParameters = Maps.newHashMap(); } public Builder withProperties(final Map<?, ?> properties, @Nullable final String prefix) { final String p = prefix == null ? "" : prefix.endsWith(".") ? prefix : prefix + "."; for (final Map.Entry<?, ?> entry : properties.entrySet()) { if (entry.getKey() != null && entry.getValue() != null && entry.getKey().toString().startsWith(p)) { final String name = entry.getKey().toString().substring(p.length()); final String value = Strings.emptyToNull(entry.getValue().toString()); if ("template".equals(name)) { withTemplate(value); } else if (name.startsWith("template.")) { withTemplateParameter(name.substring("template.".length()), value); } } } return this; } public Builder withNodeTypes(@Nullable final Iterable<? extends URI> nodeTypes) { this.nodeTypes = nodeTypes; return this; } public Builder withNodeNamespaces(@Nullable final Iterable<? extends String> nodeNamespaces) { |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ontonotes/frames/Example.java | pikes-resources | 21 |
eu/fbk/dkm/pikes/resources/util/propbank/Example.java | pikes-resources | 19 |
@XmlAccessorType(XmlAccessType.FIELD) @XmlType(name = "", propOrder = { "inflectionOrNoteOrTextOrArgOrRel" }) @XmlRootElement(name = "example") public class Example { @XmlAttribute(name = "name") @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String name; @XmlAttribute(name = "type") @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String type; @XmlAttribute(name = "src") @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String src; @XmlElements({ @XmlElement(name = "inflection", type = Inflection.class), @XmlElement(name = "note", type = Note.class), @XmlElement(name = "text", type = Text.class), @XmlElement(name = "arg", type = Arg.class), @XmlElement(name = "rel", type = Rel.class) }) protected List<Object> inflectionOrNoteOrTextOrArgOrRel; /** * Recupera il valore della proprietà name. * * @return * possible object is * {@link String } * */ public String getName() { return name; } /** * Imposta il valore della proprietà name. * * @param value * allowed object is * {@link String } * */ public void setName(String value) { this.name = value; } /** * Recupera il valore della proprietà type. * * @return * possible object is * {@link String } * */ public String getType() { return type; } /** * Imposta il valore della proprietà type. * * @param value * allowed object is * {@link String } * */ public void setType(String value) { this.type = value; } /** * Recupera il valore della proprietà src. * * @return * possible object is * {@link String } * */ public String getSrc() { return src; } /** * Imposta il valore della proprietà src. * * @param value * allowed object is * {@link String } * */ public void setSrc(String value) { this.src = value; } /** * Gets the value of the inflectionOrNoteOrTextOrArgOrRel property. * * <p> * This accessor method returns a reference to the live list, * not a snapshot. Therefore any modification you make to the * returned list will be present inside the JAXB object. * This is why there is not a <CODE>set</CODE> method for the inflectionOrNoteOrTextOrArgOrRel property. * * <p> * For example, to add a new item, do as follows: * <pre> * getInflectionOrNoteOrTextOrArgOrRel().add(newItem); * </pre> * * * <p> * Objects of the following type(s) are allowed in the list * {@link Inflection } * {@link Note } * {@link Text } * {@link Arg } * {@link Rel } * * */ public List<Object> getInflectionOrNoteOrTextOrArgOrRel() { if (inflectionOrNoteOrTextOrArgOrRel == null) { inflectionOrNoteOrTextOrArgOrRel = new ArrayList<Object>(); } return this.inflectionOrNoteOrTextOrArgOrRel; } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/tempeval/TMLtoNAF.java | pikes-resources | 26 |
eu/fbk/dkm/pikes/resources/treccani/TAOL.java | pikes-resources | 26 |
public static void main(String[] args) { try { final CommandLine cmd = CommandLine .parser() .withName("./taol-extractor") .withHeader("Convert file from Treccani XML to NAF") .withOption("i", "input", "Input folder", "FOLDER", CommandLine.Type.DIRECTORY_EXISTING, true, false, true) .withOption("o", "output", "Output folder", "FOLDER", CommandLine.Type.DIRECTORY, true, false, true) .withOption("p", "prefix", String.format("Prefix (default %s)", DEFAULT_PREFIX), "PREFIX", CommandLine.Type.STRING, true, false, false) .withLogger(LoggerFactory.getLogger("eu.fbk")).parse(args); File inputFolder = cmd.getOptionValue("input", File.class); File outputFolder = cmd.getOptionValue("output", File.class); String prefix = cmd.getOptionValue("prefix", String.class, DEFAULT_PREFIX); if (!outputFolder.exists()) { outputFolder.mkdirs(); } DocumentBuilderFactory dbFactory = DocumentBuilderFactory.newInstance(); DocumentBuilder dBuilder = dbFactory.newDocumentBuilder(); int i = 0; for (final File file : Files.fileTreeTraverser().preOrderTraversal(inputFolder)) { if (!file.isFile()) { continue; } if (file.getName().startsWith(".")) { continue; } Document doc = dBuilder.parse(file); doc.getDocumentElement().normalize(); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/trec/CR.java | pikes-resources | 70 |
eu/fbk/dkm/pikes/resources/trec/FT.java | pikes-resources | 67 |
if (file.getName().startsWith(".")) { continue; } String outputTemplate = outputDir.getAbsolutePath() + File.separator + file.getName(); File newFolder = new File(outputTemplate); newFolder.mkdirs(); outputTemplate += File.separator + "NAF"; saveFile(file, outputTemplate, urlTemplate); } } catch (Exception e) { CommandLine.fail(e); } } private static void saveFile(File inputFile, String outputFilePattern, String urlTemplate) throws IOException, SAXException, ParserConfigurationException { LOGGER.info("Input file: {}", inputFile); StringBuffer stringBuffer = new StringBuffer(); stringBuffer.append("<ROOT>\n"); stringBuffer.append(Files.toString(inputFile, Charsets.UTF_8)); stringBuffer.append("\n</ROOT>\n"); InputStream is = new ByteArrayInputStream(stringBuffer.toString().getBytes()); DocumentBuilderFactory dbFactory = DocumentBuilderFactory.newInstance(); DocumentBuilder dBuilder = dbFactory.newDocumentBuilder(); Document doc = dBuilder.parse(is); doc.getDocumentElement().normalize(); int i = 0; for (Element element : JOOX.$(doc).find("DOC")) { Element docnoElement = JOOX.$(element).find("DOCNO").get(0); Element dateElement = JOOX.$(element).find("DATE").get(0); Element headlineElement = JOOX.$(element).find("TTL").get(0); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/rdf/Renderer.java | pikes-rdf | 432 |
eu/fbk/dkm/pikes/rdf/naf/NAFRenderer.java | pikes-rdf | 373 |
public void renderObject(final Appendable out, final Object object, @Nullable final Model model) throws IOException { if (object instanceof URI) { final URI uri = (URI) object; out.append("<a>").append(shorten(uri)).append("</a>"); } else if (object instanceof Literal) { final Literal literal = (Literal) object; out.append("<span"); if (literal.getLanguage() != null) { out.append(" title=\"@").append(literal.getLanguage()).append("\""); } else if (literal.getDatatype() != null) { out.append(" title=\"").append(shorten(literal.getDatatype())).append("\""); } out.append(">").append(literal.stringValue()).append("</span>"); } else if (object instanceof BNode) { final BNode bnode = (BNode) object; out.append("_:").append(bnode.getID()); } else if (object instanceof Iterable<?>) { String separator = ""; for (final Object element : (Iterable<?>) object) { out.append(separator); renderObject(out, element, model); separator = "<br/>"; } } else if (object != null) { out.append(object.toString()); } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/trec/FT.java | pikes-resources | 67 |
eu/fbk/dkm/pikes/resources/trec/LATIMES.java | pikes-resources | 74 |
if (!file.isFile()) { continue; } String outputTemplate = outputDir.getAbsolutePath() + File.separator + file.getName(); File newFolder = new File(outputTemplate); newFolder.mkdirs(); outputTemplate += File.separator + "NAF"; saveFile(file, outputTemplate, urlTemplate); } } catch (Exception e) { CommandLine.fail(e); } } private static void saveFile(File inputFile, String outputFilePattern, String urlTemplate) throws IOException, SAXException, ParserConfigurationException { LOGGER.info("Input file: {}", inputFile); StringBuffer stringBuffer = new StringBuffer(); stringBuffer.append("<ROOT>\n"); stringBuffer.append(Files.toString(inputFile, Charsets.UTF_8)); stringBuffer.append("\n</ROOT>\n"); InputStream is = new ByteArrayInputStream(stringBuffer.toString().getBytes()); DocumentBuilderFactory dbFactory = DocumentBuilderFactory.newInstance(); DocumentBuilder dBuilder = dbFactory.newDocumentBuilder(); Document doc = dBuilder.parse(is); doc.getDocumentElement().normalize(); int i = 0; for (Element element : JOOX.$(doc).find("DOC")) { Element docnoElement = JOOX.$(element).find("DOCNO").get(0); Element dateElement = JOOX.$(element).find("DATE").get(0); Element headlineElement = JOOX.$(element).find("HEADLINE").get(0); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/tintop/annotators/raw/UKB.java | pikes-tintop | 77 |
eu/fbk/dkm/pikes/tintop/annotators/raw/UKB_MT.java | pikes-tintop | 64 |
synchronized public void run(ArrayList<HashMap<String, String>> terms) throws IOException { HashMap<String, HashMap<String, String>> backupTerms = new HashMap<>(); StringBuffer sb = new StringBuffer(); sb.append("ctx_01\n"); StringBuffer sbTokens = new StringBuffer(); int index = 0; for (HashMap<String, String> t : terms) { switch (t.get("simple_pos").toLowerCase()) { case "n": addTokenToContext(t, 'n', ++index, sbTokens, backupTerms); break; case "r": addTokenToContext(t, 'n', ++index, sbTokens, backupTerms); break; case "v": addTokenToContext(t, 'v', ++index, sbTokens, backupTerms); break; case "a": addTokenToContext(t, 'r', ++index, sbTokens, backupTerms); break; case "g": addTokenToContext(t, 'a', ++index, sbTokens, backupTerms); break; default: break; } } if (sbTokens.toString().trim().length() == 0) { return; } sb.append(sbTokens.toString()); sb.append("\n"); // Workaround to get last line to read in the output sb.append("ctx_02\n"); sb.append("be#v#workaround#1\n"); String transformedStr = sb.toString(); LOGGER.debug(transformedStr); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ontonotes/frames/Role.java | pikes-resources | 21 |
eu/fbk/dkm/pikes/resources/util/propbank/Role.java | pikes-resources | 19 |
@XmlAccessorType(XmlAccessType.FIELD) @XmlType(name = "", propOrder = { "vnrole" }) @XmlRootElement(name = "role") public class Role { @XmlAttribute(name = "n", required = true) @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String n; @XmlAttribute(name = "f") @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String f; @XmlAttribute(name = "descr", required = true) @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String descr; @XmlAttribute(name = "source") @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String source; protected List<Vnrole> vnrole; /** * Recupera il valore della proprietà n. * * @return * possible object is * {@link String } * */ public String getN() { return n; } /** * Imposta il valore della proprietà n. * * @param value * allowed object is * {@link String } * */ public void setN(String value) { this.n = value; } /** * Recupera il valore della proprietà f. * * @return * possible object is * {@link String } * */ public String getF() { return f; } /** * Imposta il valore della proprietà f. * * @param value * allowed object is * {@link String } * */ public void setF(String value) { this.f = value; } /** * Recupera il valore della proprietà descr. * * @return * possible object is * {@link String } * */ public String getDescr() { return descr; } /** * Imposta il valore della proprietà descr. * * @param value * allowed object is * {@link String } * */ public void setDescr(String value) { this.descr = value; } /** * Recupera il valore della proprietà source. * * @return * possible object is * {@link String } * */ public String getSource() { return source; } /** * Imposta il valore della proprietà source. * * @param value * allowed object is * {@link String } * */ public void setSource(String value) { this.source = value; } /** * Gets the value of the vnrole property. * * <p> * This accessor method returns a reference to the live list, * not a snapshot. Therefore any modification you make to the * returned list will be present inside the JAXB object. * This is why there is not a <CODE>set</CODE> method for the vnrole property. * * <p> * For example, to add a new item, do as follows: * <pre> * getVnrole().add(newItem); * </pre> * * * <p> * Objects of the following type(s) are allowed in the list * {@link Vnrole } * * */ public List<Vnrole> getVnrole() { if (vnrole == null) { vnrole = new ArrayList<Vnrole>(); } return this.vnrole; } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ontonotes/senses/Mappings.java | pikes-resources | 19 |
eu/fbk/dkm/pikes/resources/util/onsenses/Mappings.java | pikes-resources | 23 |
@XmlAccessorType(XmlAccessType.FIELD) @XmlType(name = "", propOrder = { "grSense", "wn", "omega", "pb", "vn", "fn" }) @XmlRootElement(name = "mappings") public class Mappings { @XmlElement(name = "gr_sense") protected String grSense; @XmlElement(required = true) protected List<Wn> wn; @XmlElement(required = true) protected String omega; @XmlElement(required = true) protected String pb; protected String vn; protected String fn; /** * Recupera il valore della proprietà grSense. * * @return * possible object is * {@link String } * */ public String getGrSense() { return grSense; } /** * Imposta il valore della proprietà grSense. * * @param value * allowed object is * {@link String } * */ public void setGrSense(String value) { this.grSense = value; } /** * Gets the value of the wn property. * * <p> * This accessor method returns a reference to the live list, * not a snapshot. Therefore any modification you make to the * returned list will be present inside the JAXB object. * This is why there is not a <CODE>set</CODE> method for the wn property. * * <p> * For example, to add a new item, do as follows: * <pre> * getWn().add(newItem); * </pre> * * * <p> * Objects of the following type(s) are allowed in the list * {@link Wn } * * */ public List<Wn> getWn() { if (wn == null) { wn = new ArrayList<Wn>(); } return this.wn; } /** * Recupera il valore della proprietà omega. * * @return * possible object is * {@link String } * */ public String getOmega() { return omega; } /** * Imposta il valore della proprietà omega. * * @param value * allowed object is * {@link String } * */ public void setOmega(String value) { this.omega = value; } /** * Recupera il valore della proprietà pb. * * @return * possible object is * {@link String } * */ public String getPb() { return pb; } /** * Imposta il valore della proprietà pb. * * @param value * allowed object is * {@link String } * */ public void setPb(String value) { this.pb = value; } /** * Recupera il valore della proprietà vn. * * @return * possible object is * {@link String } * */ public String getVn() { return vn; } /** * Imposta il valore della proprietà vn. * * @param value * allowed object is * {@link String } * */ public void setVn(String value) { this.vn = value; } /** * Recupera il valore della proprietà fn. * * @return * possible object is * {@link String } * */ public String getFn() { return fn; } /** * Imposta il valore della proprietà fn. * * @param value * allowed object is * {@link String } * */ public void setFn(String value) { this.fn = value; } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/mpqa/JohanssonAnnotator.java | pikes-resources | 203 |
eu/fbk/dkm/pikes/resources/mpqa/JohanssonAnnotator.java | pikes-resources | 289 |
opinion.setLabel("gold-eu.fbk.dkm.pikes.resources.mpqa-subjective"); Opinion.OpinionExpression opinionExpression = opinion.createOpinionExpression(expressionSpan); DataElementNode expressionNode = (DataElementNode) entity.data.children.get(0); opinionExpression.setPolarity(expressionNode.attributes.get("pol")); opinionExpression.setStrength(expressionNode.attributes.get("int")); String holderString = expressionNode.attributes.get("ns"); if (holderString != null) { String[] parts = holderString.split(","); holders: for (int i = parts.length - 1; i >= 0; i--) { String agentID = parts[i].replaceAll("[^0-9]", ""); if (hiddenAgents.containsKey(agentID)) { opinionExpression.setSentimentProductFeature(hiddenAgents.get(agentID)); break holders; } Set<Span<Term>> spans = agents.get(agentID); for (Span<Term> termSpan : spans) { Integer agentSentence = sentenceForSpan(termSpan); if (agentSentence == null) { continue; } if (!agentSentence.equals(sentence)) { continue; } if (termSpan == null) { continue; } holderSpan = termSpan; break holders; } } } if (holderSpan.size() > 0) { Opinion.OpinionHolder opinionHolder = opinion.createOpinionHolder(holderSpan); } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/trec/Queries.java | pikes-resources | 70 |
eu/fbk/dkm/pikes/resources/trec/QueriesTSV.java | pikes-resources | 58 |
String content = FileUtils.readFileToString(inputfile, Charsets.UTF_8); StringBuffer newContent = new StringBuffer(); newContent.append("<root>\n"); newContent.append(content .replaceAll("<title>", "</num>\n<title>") .replaceAll("<desc>", "</title>\n<desc>") .replaceAll("<narr>", "</desc>\n<narr>") .replaceAll("</top>", "</narr>\n</top>") .replaceAll("R&D", "R&D") ); newContent.append("</root>\n"); Document doc = dBuilder.parse(new ByteArrayInputStream(newContent.toString().getBytes(Charsets.UTF_8))); for (Element element : JOOX.$(doc).find("top")) { Element numElement = JOOX.$(element).find("num").get(0); Element titleElement = JOOX.$(element).find("title").get(0); Element descElement = JOOX.$(element).find("desc").get(0); String number = "q" + numElement.getTextContent().trim().substring(7).trim(); String title = titleElement.getTextContent().trim().replaceAll("\\s+", " "); String desc = descElement.getTextContent().trim().substring(12).trim().replaceAll("\\s+", " "); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/mpqa/CorpusPreprocessor.java | pikes-resources | 402 |
eu/fbk/dkm/pikes/resources/mpqa/CorpusPreprocessor.java | pikes-resources | 446 |
fields.put("type", "expressive"); String id = thisRecord.getValue("id"); if (id != null) { fields.put("id", id); } final Span expressionSpan = thisRecord.getSpan().align(text); fields.put("expression", expressionSpan.toString()); // Sentence Span sentenceSpan = getSentenceSpan(thisRecord, sentences, fields, text, documentURI); if (sentenceSpan == null) { continue; } for (final String name : DSA_FIELDS) { String value = thisRecord.getValue(name); if (value != null) { fields.put("dsa-" + name, value); } } // Holder String sources = thisRecord.getValue("nested-source"); addSources(sources, agentRecords, lastRecords, sentenceSpan, fields, documentURI, text); // Emit TSV string tsvWriter.append(getTsvString(documentURI, fields)); // Print debug LOGGER.debug(fields.get("type").toString()); LOGGER.debug(fields.toString()); for (String expression : fields.get("expression")) { LOGGER.debug(expression); Span span = new Span(expression); LOGGER.debug(span.apply(text)); } System.out.println(); } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/rdf/Renderer.java | pikes-rdf | 112 |
eu/fbk/dkm/pikes/rdf/naf/NAFRenderer.java | pikes-rdf | 477 |
private static final Mustache loadTemplate(final Object spec) { Preconditions.checkNotNull(spec); try { if (spec instanceof Mustache) { return (Mustache) spec; } final DefaultMustacheFactory factory = new DefaultMustacheFactory(); // factory.setExecutorService(Environment.getPool()); // BROKEN URL url = spec instanceof URL ? (URL) spec : null; if (url == null) { try { url = Renderer.class.getResource(spec.toString()); } catch (final Throwable ex) { // ignore } } if (url == null) { final File file = spec instanceof File ? (File) spec : new File(spec.toString()); if (file.exists()) { url = file.toURI().toURL(); } } if (url != null) { return factory.compile(new InputStreamReader(url.openStream(), Charsets.UTF_8), url.toString()); } else { return factory.compile(new StringReader(spec.toString()), Hash.murmur3(spec.toString()).toString()); } } catch (final IOException ex) { throw new IllegalArgumentException("Could not create Mustache template for " + spec); } } |
File | Project | Line |
---|---|---|
ixa/kaflib/Opinion.java | pikes-naflib | 93 |
ixa/kaflib/Opinion.java | pikes-naflib | 210 |
ixa/kaflib/Opinion.java | pikes-naflib | 397 |
} public List<Term> getTerms() { return this.span.getTargets(); } public void addTerm(final Term term) { this.span.addTarget(term); } public void addTerm(final Term term, final boolean isHead) { this.span.addTarget(term, isHead); } public Span<Term> getSpan() { return this.span; } public void setSpan(final Span<Term> span) { this.span = span; } public ExternalRef getExternalRef(final String resource) { for (final ExternalRef ref : this.externalReferences) { if (ref.getResource().equalsIgnoreCase(resource)) { return ref; } } return null; } public List<ExternalRef> getExternalRefs() { return this.externalReferences; } public void addExternalRef(final ExternalRef externalRef) { this.externalReferences.add(externalRef); } public void addExternalRefs(final List<ExternalRef> externalRefs) { this.externalReferences.addAll(externalRefs); } @Override public boolean equals(final Object object) { if (object == this) { return true; } if (!(object instanceof OpinionTarget)) { |
File | Project | Line |
---|---|---|
ixa/kaflib/Entity.java | pikes-naflib | 141 |
ixa/kaflib/Feature.java | pikes-naflib | 112 |
} public List<ExternalRef> getExternalRefs() { return externalReferences; } public void addExternalRef(ExternalRef externalRef) { externalReferences.add(externalRef); } public void addExternalRefs(List<ExternalRef> externalRefs) { externalReferences.addAll(externalRefs); } public String getSpanStr(Span<Term> span) { String str = ""; for (Term term : span.getTargets()) { if (!str.isEmpty()) { str += " "; } str += term.getStr(); } return str; } public String getStr() { return getSpanStr(this.getSpans().get(0)); } /** Deprecated */ public List<List<Term>> getReferences() { List<List<Term>> list = new ArrayList<List<Term>>(); for (Span<Term> span : this.references) { list.add(span.getTargets()); } return list; } /** Deprecated */ public void addReference(List<Term> span) { this.references.add(KAFDocument.<Term>list2Span(span)); } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ecb/ECBevaluator.java | pikes-resources | 205 |
eu/fbk/dkm/pikes/resources/ecb/ECBevaluator2.java | pikes-resources | 123 |
baselineWriter.close(); // Set<Set> goldClusters = new HashSet<>(); // for (String key : goldTmpClusters.keySet()) { // Set<String> cluster = goldTmpClusters.get(key); // if (cluster.size() > 1 || !removeAloneClusters) { // goldClusters.add(cluster); // } // } // LOGGER.info("Gold clusters: {}", goldClusters.size()); in = new FileReader(inputCsv); records = CSVFormat.EXCEL.withHeader().parse(in); // Size must be always 4! int clusterID = 0; HashMap<String, Integer> clusterIndexes = new HashMap<>(); HashMultimap<Integer, String> tmpClusters = HashMultimap.create(); for (CSVRecord record : records) { Matcher matcher; String id1 = null; String id2 = null; matcher = tokenPattern.matcher(record.get(1)); if (matcher.find()) { id1 = matcher.group(1) + "_" + matcher.group(2) + "_" + matcher.group(3); } matcher = tokenPattern.matcher(record.get(3)); if (matcher.find()) { id2 = matcher.group(1) + "_" + matcher.group(2) + "_" + matcher.group(3); } Integer index1 = clusterIndexes.get(id1); Integer index2 = clusterIndexes.get(id2); if (index1 == null && index2 == null) { clusterID++; if (!checkGold || okEvents.contains(id2)) { |
File | Project | Line |
---|---|---|
eu/fbk/shell/mdfsa/data/structures/DomainGraph.java | pikes-raid | 425 |
eu/fbk/shell/mdfsa/data/structures/DomainGraph.java | pikes-raid | 506 |
double deadzone = this.deadzone; int judged = 0; double precision = 0.0; double recall = 0.0; for(DatasetInstance di: instances) { ArrayList<String> features = di.getFeatures(); int testPolarity = di.getPolarity(); double inferredPolarity = 0.0; double a = 0.0; double b = 0.0; double c = 0.0; double d = 0.0; int mappedFeatures = 0; for(String currentFeature: features) { //Long featureId = labels.get(currentFeature); ArrayList<Long> featureIds = this.graph.getFeatureIds(currentFeature); if(featureIds != null) { for(Long featureId: featureIds) { mappedFeatures++; FuzzyMembership fm = this.polarities.get(featureId); if(fm != null) { inferredPolarity += fm.getCentroid(); a += fm.getA(); b += fm.getB(); c += fm.getC(); d += fm.getD(); } } } } a /= (double) mappedFeatures; b /= (double) mappedFeatures; c /= (double) mappedFeatures; d /= (double) mappedFeatures; //inferredPolarity /= (double) mappedFeatures; inferredPolarity = (b + c) / 2.0; |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/rdf/RDFGenerator.java | pikes-rdf | 1978 |
eu/fbk/dkm/pikes/rdf/util/ProcessorASNorm.java | pikes-rdf | 63 |
} private URI hash(final Resource subject, final URI predicate, final Value object) { final List<String> list = Lists.newArrayList(); for (final Value value : new Value[] { subject, predicate, object }) { if (value instanceof URI) { list.add("\u0001"); list.add(value.stringValue()); } else if (value instanceof BNode) { list.add("\u0002"); list.add(((BNode) value).getID()); } else if (value instanceof Literal) { final Literal l = (Literal) value; list.add("\u0003"); list.add(l.getLabel()); if (l.getDatatype() != null) { list.add(l.getDatatype().stringValue()); } else if (l.getLanguage() != null) { list.add(l.getLanguage()); } } } final String id = Hash.murmur3(list.toArray(new String[list.size()])).toString(); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ontonotes/senses/Inventory.java | pikes-resources | 21 |
eu/fbk/dkm/pikes/resources/util/onsenses/Inventory.java | pikes-resources | 26 |
@XmlAccessorType(XmlAccessType.FIELD) @XmlType(name = "", propOrder = { "commentary", "sense", "wordmeta" }) @XmlRootElement(name = "inventory") public class Inventory { @XmlAttribute(name = "lemma", required = true) @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String lemma; protected String commentary; @XmlElement(required = true) protected List<Sense> sense; @XmlElement(name = "WORD_META", required = true) protected WORDMETA wordmeta; /** * Recupera il valore della proprietà lemma. * * @return * possible object is * {@link String } * */ public String getLemma() { return lemma; } /** * Imposta il valore della proprietà lemma. * * @param value * allowed object is * {@link String } * */ public void setLemma(String value) { this.lemma = value; } /** * Recupera il valore della proprietà commentary. * * @return * possible object is * {@link String } * */ public String getCommentary() { return commentary; } /** * Imposta il valore della proprietà commentary. * * @param value * allowed object is * {@link String } * */ public void setCommentary(String value) { this.commentary = value; } /** * Gets the value of the sense property. * * <p> * This accessor method returns a reference to the live list, * not a snapshot. Therefore any modification you make to the * returned list will be present inside the JAXB object. * This is why there is not a <CODE>set</CODE> method for the sense property. * * <p> * For example, to add a new item, do as follows: * <pre> * getSense().add(newItem); * </pre> * * * <p> * Objects of the following type(s) are allowed in the list * {@link Sense } * * */ public List<Sense> getSense() { if (sense == null) { sense = new ArrayList<Sense>(); } return this.sense; } /** * Recupera il valore della proprietà wordmeta. * * @return * possible object is * {@link WORDMETA } * */ public WORDMETA getWORDMETA() { return wordmeta; } /** * Imposta il valore della proprietà wordmeta. * * @param value * allowed object is * {@link WORDMETA } * */ public void setWORDMETA(WORDMETA value) { this.wordmeta = value; } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/trec/CR.java | pikes-resources | 45 |
eu/fbk/dkm/pikes/resources/trec/FBIS.java | pikes-resources | 50 |
eu/fbk/dkm/pikes/resources/trec/FR94.java | pikes-resources | 47 |
eu/fbk/dkm/pikes/resources/trec/FT.java | pikes-resources | 45 |
eu/fbk/dkm/pikes/resources/trec/LATIMES.java | pikes-resources | 49 |
.withHeader("Extract CR documents from TREC dataset and save them in NAF format") .withOption("i", "input", "Input folder", "FOLDER", CommandLine.Type.DIRECTORY_EXISTING, true, false, true) .withOption("o", "output", "Output folder", "FOLDER", CommandLine.Type.DIRECTORY, true, false, true) .withOption("u", "url-template", "URL template (with %d for the ID)", "URL", CommandLine.Type.STRING, true, false, false) .withLogger(LoggerFactory.getLogger("eu.fbk")) // .parse(args); File inputDir = cmd.getOptionValue("input", File.class); String urlTemplate = DEFAULT_URL; if (cmd.hasOption("url-template")) { urlTemplate = cmd.getOptionValue("url-template", String.class); } File outputDir = cmd.getOptionValue("output", File.class); if (!outputDir.exists()) { outputDir.mkdirs(); } for (final File file : Files.fileTreeTraverser().preOrderTraversal(inputDir)) { if (!file.isFile()) { continue; } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/mpqa/JohanssonAnnotator.java | pikes-resources | 294 |
eu/fbk/dkm/pikes/resources/mpqa/JohanssonAnnotator.java | pikes-resources | 353 |
opinionExpression.setStrength(expressionNode.attributes.get("int")); String holderString = expressionNode.attributes.get("ns"); if (holderString != null) { String[] parts = holderString.split(","); holders: for (int i = parts.length - 1; i >= 0; i--) { String agentID = parts[i].replaceAll("[^0-9]", ""); if (hiddenAgents.containsKey(agentID)) { opinionExpression.setSentimentProductFeature(hiddenAgents.get(agentID)); break holders; } Set<Span<Term>> spans = agents.get(agentID); for (Span<Term> termSpan : spans) { Integer agentSentence = sentenceForSpan(termSpan); if (agentSentence == null) { continue; } if (!agentSentence.equals(sentence)) { continue; } if (termSpan == null) { continue; } holderSpan = termSpan; break holders; } } } if (holderSpan.size() > 0) { Opinion.OpinionHolder opinionHolder = opinion.createOpinionHolder(holderSpan); } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/mpqa/JohanssonAnnotator.java | pikes-resources | 208 |
eu/fbk/dkm/pikes/resources/mpqa/JohanssonAnnotator.java | pikes-resources | 353 |
opinionExpression.setStrength(expressionNode.attributes.get("int")); String holderString = expressionNode.attributes.get("ns"); if (holderString != null) { String[] parts = holderString.split(","); holders: for (int i = parts.length - 1; i >= 0; i--) { String agentID = parts[i].replaceAll("[^0-9]", ""); if (hiddenAgents.containsKey(agentID)) { opinionExpression.setSentimentProductFeature(hiddenAgents.get(agentID)); break holders; } Set<Span<Term>> spans = agents.get(agentID); for (Span<Term> termSpan : spans) { Integer agentSentence = sentenceForSpan(termSpan); if (agentSentence == null) { continue; } if (!agentSentence.equals(sentence)) { continue; } if (termSpan == null) { continue; } holderSpan = termSpan; break holders; } } } if (holderSpan.size() > 0) { Opinion.OpinionHolder opinionHolder = opinion.createOpinionHolder(holderSpan); } |
File | Project | Line |
---|---|---|
eu/fbk/shell/mdfsa/data/structures/DomainGraph.java | pikes-raid | 873 |
eu/fbk/shell/mdfsa/data/structures/DomainGraph.java | pikes-raid | 901 |
new FileInputStream(modelPath + currentDataset + "." + type + ".mdfsa")); SerializableDomainGraph sdg = (SerializableDomainGraph) objectInputStream.readObject(); this.prp = sdg.getPrp(); this.graph = sdg.getGraph(); this.polarities = sdg.getPolarities(); this.conceptsConvergenceIterationsValues = sdg.getConceptsConvergenceIterationsValues(); this.startPolarities = sdg.getStartPolarities(); this.tempInDomainStartPolarities = sdg.getTempInDomainStartPolarities(); this.tempOutDomainStartPolarities = sdg.getTempOutDomainStartPolarities(); this.currentPolarities = sdg.getCurrentPolarities(); this.tokensCounter = sdg.getTokensCounter(); this.inDomainTokensCounter = sdg.getInDomainTokensCounter(); this.outDomainTokensCounter = sdg.getOutDomainTokensCounter(); this.currentGraphConvergenceValue = sdg.getCurrentGraphConvergenceValue(); this.currentAveragePolarity = sdg.getCurrentAveragePolarity(); this.iteration = sdg.getIteration(); this.propagationRate = sdg.getPropagationRate(); this.convergenceLimit = sdg.getConvergenceLimit(); this.deadzone = sdg.getDeadzone(); this.annealingRate = sdg.getAnnealingRate(); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ecb/ECBPlusEvaluator.java | pikes-resources | 279 |
eu/fbk/dkm/pikes/resources/ecb/ECBPlusEvaluatorLemma.java | pikes-resources | 279 |
eu/fbk/dkm/pikes/resources/ecb/ECBevaluator.java | pikes-resources | 223 |
HashMultimap<Integer, String> theClusters = HashMultimap.create(); for (CSVRecord record : records) { Matcher matcher; String id1 = null; String id2 = null; matcher = tokenPattern.matcher(record.get(1)); if (matcher.find()) { id1 = matcher.group(1) + "_" + matcher.group(2) + "_" + matcher.group(3); } matcher = tokenPattern.matcher(record.get(3)); if (matcher.find()) { id2 = matcher.group(1) + "_" + matcher.group(2) + "_" + matcher.group(3); } // System.out.println(id1); // System.out.println(id2); Integer index1 = clusterIndexes.get(id1); Integer index2 = clusterIndexes.get(id2); // System.out.println(index1); // System.out.println(index2); if (index1 == null && index2 == null) { clusterID++; if (!checkGold || okEvents.contains(id2)) { if (id2 != null) { |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/raid/pipeline/PipelineTrainer.java | pikes-raid | 72 |
eu/fbk/dkm/pikes/raid/sbrs/SBRSTrainer.java | pikes-raid | 47 |
this.jointSpan = Boolean.parseBoolean(properties.getProperty("joint", "false")); this.holderLinkTrainer = hasHolder ? LinkLabeller.train("NN", "PRP", "JJP", "DTP", "WP") : null; this.targetLinkTrainer = hasTarget ? LinkLabeller.train("NN", "PRP", "JJP", "DTP", "WP", "VB") : null; if (this.jointSpan) { final SpanLabeller.Trainer t = hasHolder || hasTarget ? SpanLabeller.train() : null; this.holderSpanTrainer = t; this.targetSpanTrainer = t; } else { this.holderSpanTrainer = hasHolder ? SpanLabeller.train() : null; this.targetSpanTrainer = hasTarget ? SpanLabeller.train() : null; } this.filter = NAFFilter.builder(false).withTermSenseCompletion(true) .withEntityAddition(true).withEntityRemoveOverlaps(true) .withEntitySpanFixing(true).withSRLPredicateAddition(true) .withSRLRemoveWrongRefs(true).withSRLSelfArgFixing(true).build(); } @Override protected void doFilter(final KAFDocument document) { |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ecb/ECBparser.java | pikes-resources | 32 |
eu/fbk/dkm/pikes/resources/tempeval/TMLtoNAF.java | pikes-resources | 31 |
eu/fbk/dkm/pikes/resources/treccani/TAOL.java | pikes-resources | 31 |
.withHeader("Convert file from ecb annotated txt to NAF") .withOption("i", "input", "Input folder", "FOLDER", CommandLine.Type.DIRECTORY_EXISTING, true, false, true) .withOption("o", "output", "Output folder", "FOLDER", CommandLine.Type.DIRECTORY, true, false, true) .withOption("p", "prefix", String.format("Prefix (default %s)", DEFAULT_PREFIX), "PREFIX", CommandLine.Type.STRING, true, false, false) .withLogger(LoggerFactory.getLogger("eu.fbk")).parse(args); File inputFolder = cmd.getOptionValue("input", File.class); File outputFolder = cmd.getOptionValue("output", File.class); String prefix = cmd.getOptionValue("prefix", String.class, DEFAULT_PREFIX); if (!outputFolder.exists()) { outputFolder.mkdirs(); } DocumentBuilderFactory dbFactory = DocumentBuilderFactory.newInstance(); DocumentBuilder dBuilder = dbFactory.newDocumentBuilder(); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ontonotes/frames/ObjectFactory.java | pikes-resources | 28 |
eu/fbk/dkm/pikes/resources/util/propbank/ObjectFactory.java | pikes-resources | 25 |
@XmlRegistry public class ObjectFactory { /** * Create a new ObjectFactory that can be used to create new instances of schema derived classes for package: eu.fbk.dkm.pikes.resources.ontonotes.frames * */ public ObjectFactory() { } /** * Create an instance of {@link Note } * */ public Note createNote() { return new Note(); } /** * Create an instance of {@link Predicate } * */ public Predicate createPredicate() { return new Predicate(); } /** * Create an instance of {@link Roleset } * */ public Roleset createRoleset() { return new Roleset(); } /** * Create an instance of {@link Vnrole } * */ public Vnrole createVnrole() { return new Vnrole(); } /** * Create an instance of {@link Role } * */ public Role createRole() { return new Role(); } /** * Create an instance of {@link Frameset } * */ public Frameset createFrameset() { return new Frameset(); } /** * Create an instance of {@link Roles } * */ public Roles createRoles() { return new Roles(); } /** * Create an instance of {@link Arg } * */ public Arg createArg() { return new Arg(); } /** * Create an instance of {@link Rel } * */ public Rel createRel() { return new Rel(); } /** * Create an instance of {@link Text } * */ public Text createText() { return new Text(); } /** * Create an instance of {@link Example } * */ public Example createExample() { return new Example(); } /** * Create an instance of {@link Inflection } * */ public Inflection createInflection() { return new Inflection(); } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ecb/ECBPlusEvaluator.java | pikes-resources | 279 |
eu/fbk/dkm/pikes/resources/ecb/ECBPlusEvaluatorLemma.java | pikes-resources | 279 |
eu/fbk/dkm/pikes/resources/ecb/ECBevaluator2.java | pikes-resources | 131 |
HashMultimap<Integer, String> theClusters = HashMultimap.create(); for (CSVRecord record : records) { Matcher matcher; String id1 = null; String id2 = null; matcher = tokenPattern.matcher(record.get(1)); if (matcher.find()) { id1 = matcher.group(1) + "_" + matcher.group(2) + "_" + matcher.group(3); } matcher = tokenPattern.matcher(record.get(3)); if (matcher.find()) { id2 = matcher.group(1) + "_" + matcher.group(2) + "_" + matcher.group(3); } // System.out.println(id1); // System.out.println(id2); Integer index1 = clusterIndexes.get(id1); Integer index2 = clusterIndexes.get(id2); // System.out.println(index1); // System.out.println(index2); if (index1 == null && index2 == null) { clusterID++; if (!checkGold || okEvents.contains(id2)) { |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/tintop/server/EverythingHandler.java | pikes-tintop | 65 |
eu/fbk/dkm/pikes/tintop/server/TriplesHandler.java | pikes-tintop | 67 |
try { HashMap<String, Object> demoProperties = new HashMap<>(); demoProperties.put("renderer.template.title", "PIKES demo"); if (backLink) { demoProperties.put("renderer.template.backlink", "javascript:history.back();"); } else { demoProperties.put("renderer.template.backlink", pipeline.getDefaultConfig().getProperty("back_alt_link")); demoProperties.put("renderer.template.backlabel", pipeline.getDefaultConfig().getProperty("back_alt_text")); } boolean fusion = request.getParameter("rdf_fusion") != null; boolean normalization = request.getParameter("rdf_compaction") != null; demoProperties.put("generator.fusion", fusion); demoProperties.put("generator.normalization", normalization); NAFFilter filter = NAFFilter.builder().withProperties(pipeline.getDefaultConfig(), "filter").build(); RDFGenerator generator = RDFGenerator.builder().withProperties(demoProperties, "generator").build(); |
File | Project | Line |
---|---|---|
ixa/kaflib/Entity.java | pikes-naflib | 56 |
ixa/kaflib/Feature.java | pikes-naflib | 41 |
for (Span<Term> span : entity.getSpans()) { /* Copy span */ List<Term> targets = span.getTargets(); List<Term> copiedTargets = new ArrayList<Term>(); for (Term term : targets) { Term copiedTerm = terms.get(term.getId()); if (copiedTerm == null) { throw new IllegalStateException("Term not found when copying " + id); } copiedTargets.add(copiedTerm); } if (span.hasHead()) { Term copiedHead = terms.get(span.getHead().getId()); this.references.add(new Span<Term>(copiedTargets, copiedHead)); } else { this.references.add(new Span<Term>(copiedTargets)); } } /* Copy external references */ this.externalReferences = new ArrayList<ExternalRef>(); for (ExternalRef externalRef : entity.getExternalRefs()) { |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/reader/LKCollectionReader.java | pikes-resources | 650 |
eu/fbk/dkm/pikes/resources/reader/LKCollectionReader.java | pikes-resources | 697 |
private <T> ArrayList<T> tsort(ListMap<T, T> depGraph) { HashSet<T> starts = new HashSet<T>(); for (T k : depGraph.keySet()) { starts.add(k); } for (T k : depGraph.keySet()) { starts.removeAll(depGraph.get(k)); } if (starts.size() == 0) { throw new RuntimeException("cyclic or empty graph!"); } LinkedList<T> q = new LinkedList(starts); ArrayList<T> out = new ArrayList(); while (!q.isEmpty()) { T t = q.removeFirst(); out.add(t); ArrayList<T> sl = depGraph.get(t); if (sl != null) { |
File | Project | Line |
---|---|---|
ixa/kaflib/Mark.java | pikes-naflib | 54 |
ixa/kaflib/Term.java | pikes-naflib | 343 |
this.sid = id; } public boolean hasType() { return type != null; } public String getType() { return type; } public void setType(String type) { this.type = type; } public boolean hasLemma() { return lemma != null; } public String getLemma() { return lemma; } public void setLemma(String lemma) { this.lemma = lemma; } public boolean hasPos() { return pos != null; } public String getPos() { return pos; } public void setPos(String pos) { this.pos = pos; } public boolean hasMorphofeat() { return morphofeat != null; } public String getMorphofeat() { return morphofeat; } public void setMorphofeat(String morphofeat) { this.morphofeat = morphofeat; } public boolean hasCase() { return markcase != null; |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ontonotes/frames/Predicate.java | pikes-resources | 21 |
eu/fbk/dkm/pikes/resources/util/propbank/Predicate.java | pikes-resources | 19 |
@XmlAccessorType(XmlAccessType.FIELD) @XmlType(name = "", propOrder = { "noteOrRoleset" }) @XmlRootElement(name = "predicate") public class Predicate { @XmlAttribute(name = "lemma", required = true) @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String lemma; @XmlElements({ @XmlElement(name = "note", type = Note.class), @XmlElement(name = "roleset", type = Roleset.class) }) protected List<Object> noteOrRoleset; /** * Recupera il valore della proprietà lemma. * * @return * possible object is * {@link String } * */ public String getLemma() { return lemma; } /** * Imposta il valore della proprietà lemma. * * @param value * allowed object is * {@link String } * */ public void setLemma(String value) { this.lemma = value; } /** * Gets the value of the noteOrRoleset property. * * <p> * This accessor method returns a reference to the live list, * not a snapshot. Therefore any modification you make to the * returned list will be present inside the JAXB object. * This is why there is not a <CODE>set</CODE> method for the noteOrRoleset property. * * <p> * For example, to add a new item, do as follows: * <pre> * getNoteOrRoleset().add(newItem); * </pre> * * * <p> * Objects of the following type(s) are allowed in the list * {@link Note } * {@link Roleset } * * */ public List<Object> getNoteOrRoleset() { if (noteOrRoleset == null) { noteOrRoleset = new ArrayList<Object>(); } return this.noteOrRoleset; } } |
File | Project | Line |
---|---|---|
eu/fbk/shell/mdfsa/data/structures/SentenceStructuredRepresentation.java | pikes-raid | 132 |
eu/fbk/shell/mdfsa/data/structures/SentenceStructuredRepresentation.java | pikes-raid | 198 |
this.semanticConcepts = new ArrayList<String>(); //String terms = this.lexString.replaceAll("\\./\\.", ""); //String[] termsList = terms.split(" "); String[] termsList = this.lexString.split(" "); boolean compoundNounFlag = false; for(String currentTerm : termsList) { String[] atom = currentTerm.split("/"); if(atom.length > 1) { Integer stopFlag = wnl.getStopwords().get(atom[0]); if(stopFlag != null) continue; if(atom[1].compareTo("NN") == 0 || atom[1].compareTo("NNP") == 0 || atom[1].compareTo("NNPS") == 0 || atom[1].compareTo("NNS") == 0 || atom[1].compareTo("FW") == 0) { String newAspect; if(compoundNounFlag == true) { newAspect = this.semanticConcepts.get(this.semanticConcepts.size() - 1); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/raid/mdfsa/parser/DatasetInstanceParser.java | pikes-raid | 117 |
eu/fbk/dkm/pikes/raid/mdfsa/parser/ReviewsParser.java | pikes-raid | 169 |
} /** * Utility method that convert the blitzer review in the eswc2014 challenge format */ public void convertReviewToESWCChallenge(String filename, String datasetName) { FileManager fm = new FileManager(filename, Mode.READ); //String content = fm.importFullTextContent(); ArrayList<String> contents = fm.importSimpleTextContent(); FileManager rdfOut = new FileManager("/home/drago/Documents/java_projects/research/nlp/multi_domain_fuzzy_sentiment_analysis/eswc2014_challenge_mdfsa_dragoni/task3/" + datasetName + ".validation.rdf.xml", Mode.WRITE); rdfOut.write("<?xml version=\"1.0\" encoding=\"UTF-8\"?>"); rdfOut.write("<rdf:RDF xmlns:rdf=\"http://www.w3.org/1999/02/22-rdf-syntax-ns#\">"); int reviewId = 1; //int startText = content.indexOf("<review_text"); //int endText = content.indexOf("</review_text", startText + 10); //while (startText != -1) { for(String currentReview: contents) { /* Gets the review text */ //String currentReview = content.substring(startText + 14, endText - 1).replaceAll("\n", ""); currentReview = currentReview.replace("&", "&"); //currentReview = "I gave up to go to supermarkets yesterday."; /* Write the review in the RDF format */ rdfOut.write("\t<rdf:Description rdf:about=\"http://sentic.net/challenge/sentence_" + reviewId + "\">"); //rdfOut.write("\t\t<sentence xmlns=\"http://sentic.net/challenge/\" rdf:resource=\"http://sentic.net/challenge/sentence_" + reviewId + "\">"); rdfOut.write("\t\t\t<text xmlns=\"http://sentic.net/challenge/\" rdf:datatype=\"http://www.w3.org/TR/rdf-text/\">"); //rdfOut.write("\t\t\t<![CDATA[" + currentReview + "]]>"); rdfOut.write("\t\t\t" + currentReview + ""); rdfOut.write("\t\t\t</text>"); //rdfOut.write("\t\t</sentence>"); rdfOut.write("\t</rdf:Description>"); /* Gets next text to analyze */ //startText = content.indexOf("<review_text", endText + 10); //endText = content.indexOf("</review_text", startText + 10); //System.out.println(reviewId + " - " + startText + " - " + endText); reviewId++; } rdfOut.write("</rdf:RDF>"); rdfOut.close(); fm.close(); } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ontonotes/frames/Arg.java | pikes-resources | 19 |
eu/fbk/dkm/pikes/resources/util/propbank/Arg.java | pikes-resources | 17 |
@XmlAccessorType(XmlAccessType.FIELD) @XmlType(name = "", propOrder = { "value" }) @XmlRootElement(name = "arg") public class Arg { @XmlAttribute(name = "n", required = true) @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String n; @XmlAttribute(name = "f") @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String f; @XmlValue protected String value; /** * Recupera il valore della proprietà n. * * @return * possible object is * {@link String } * */ public String getN() { return n; } /** * Imposta il valore della proprietà n. * * @param value * allowed object is * {@link String } * */ public void setN(String value) { this.n = value; } /** * Recupera il valore della proprietà f. * * @return * possible object is * {@link String } * */ public String getF() { return f; } /** * Imposta il valore della proprietà f. * * @param value * allowed object is * {@link String } * */ public void setF(String value) { this.f = value; } /** * Recupera il valore della proprietà value. * * @return * possible object is * {@link String } * */ public String getvalue() { return value; } /** * Imposta il valore della proprietà value. * * @param value * allowed object is * {@link String } * */ public void setvalue(String value) { this.value = value; } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ontonotes/senses/Wn.java | pikes-resources | 19 |
eu/fbk/dkm/pikes/resources/util/onsenses/Wn.java | pikes-resources | 24 |
@XmlAccessorType(XmlAccessType.FIELD) @XmlType(name = "", propOrder = { "value" }) @XmlRootElement(name = "wn") public class Wn { @XmlAttribute(name = "version", required = true) @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String version; @XmlAttribute(name = "lemma") @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String lemma; @XmlValue protected String value; /** * Recupera il valore della proprietà version. * * @return * possible object is * {@link String } * */ public String getVersion() { return version; } /** * Imposta il valore della proprietà version. * * @param value * allowed object is * {@link String } * */ public void setVersion(String value) { this.version = value; } /** * Recupera il valore della proprietà lemma. * * @return * possible object is * {@link String } * */ public String getLemma() { return lemma; } /** * Imposta il valore della proprietà lemma. * * @param value * allowed object is * {@link String } * */ public void setLemma(String value) { this.lemma = value; } /** * Recupera il valore della proprietà value. * * @return * possible object is * {@link String } * */ public String getvalue() { return value; } /** * Imposta il valore della proprietà value. * * @param value * allowed object is * {@link String } * */ public void setvalue(String value) { this.value = value; } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/tintop/annotators/raw/UKB.java | pikes-tintop | 138 |
eu/fbk/dkm/pikes/tintop/annotators/raw/UKB_MT.java | pikes-tintop | 142 |
String line; while ((line = brCleanUp.readLine()) != null) { LOGGER.trace(line); String[] parts = line.split("\\s+"); String context = parts[0]; if (context.equals("ctx_02")) { break; } String tokenID = parts[1]; HashMap<String, String> thisTerm = backupTerms.get(tokenID); if (thisTerm == null) { continue; } for (int i = 2; i < parts.length; i++) { String[] scores = parts[i].split("/"); if (scores.length < 2) { break; } String wn = scores[0]; thisTerm.put("wordnet", wn); break; } } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ecb/ECBPlusEvaluator.java | pikes-resources | 168 |
eu/fbk/dkm/pikes/resources/ecb/ECBPlusEvaluatorLemma.java | pikes-resources | 168 |
eu/fbk/dkm/pikes/resources/ecb/ECBevaluator.java | pikes-resources | 132 |
if (FOLDER != null && !folder.equals(FOLDER)) { continue; } LOGGER.debug(file.getAbsolutePath()); KAFDocument document = KAFDocument.createFromFile(file); for (Coref coref : document.getCorefs()) { if (coref.getType() == null) { continue; } if (!coref.getType().equals("event-gold")) { continue; } Integer cluster = Integer.parseInt(coref.getCluster()); String idCluster = String.valueOf(1000 * folder + cluster); for (Span<Term> termSpan : coref.getSpans()) { Term term = termSpan.getTargets().get(0); String lemma = term.getLemma(); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/rdf/Renderer.java | pikes-rdf | 567 |
eu/fbk/dkm/pikes/rdf/naf/NAFRenderer.java | pikes-rdf | 540 |
eu/fbk/dkm/pikes/resources/NAFFilter.java | pikes-resources | 1804 |
} public Builder withProperties(final Map<?, ?> properties, @Nullable final String prefix) { final String p = prefix == null ? "" : prefix.endsWith(".") ? prefix : prefix + "."; for (final Map.Entry<?, ?> entry : properties.entrySet()) { if (entry.getKey() != null && entry.getValue() != null && entry.getKey().toString().startsWith(p)) { final String name = entry.getKey().toString().substring(p.length()); final String value = Strings.emptyToNull(entry.getValue().toString()); if ("template".equals(name)) { |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/rdf/RDFGenerator.java | pikes-rdf | 239 |
eu/fbk/dkm/pikes/rdf/Renderer.java | pikes-rdf | 569 |
eu/fbk/dkm/pikes/rdf/naf/NAFRenderer.java | pikes-rdf | 542 |
eu/fbk/dkm/pikes/resources/NAFFilter.java | pikes-resources | 1815 |
public Builder withProperties(final Map<?, ?> properties, @Nullable final String prefix) { final String p = prefix == null ? "" : prefix.endsWith(".") ? prefix : prefix + "."; for (final Map.Entry<?, ?> entry : properties.entrySet()) { if (entry.getKey() != null && entry.getValue() != null && entry.getKey().toString().startsWith(p)) { final String name = entry.getKey().toString().substring(p.length()); final String value = Strings.emptyToNull(entry.getValue().toString()); if ("fusion".equals(name)) { |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/rdf/RDFGenerator.java | pikes-rdf | 1983 |
eu/fbk/dkm/pikes/rdf/util/ProcessorASNorm.java | pikes-rdf | 68 |
eu/fbk/dkm/pikes/rdf/util/ProcessorASNorm.java | pikes-rdf | 98 |
if (value instanceof URI) { list.add("\u0001"); list.add(value.stringValue()); } else if (value instanceof BNode) { list.add("\u0002"); list.add(((BNode) value).getID()); } else if (value instanceof Literal) { final Literal l = (Literal) value; list.add("\u0003"); list.add(l.getLabel()); if (l.getDatatype() != null) { list.add(l.getDatatype().stringValue()); } else if (l.getLanguage() != null) { list.add(l.getLanguage()); } } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/trec/FBIS.java | pikes-resources | 179 |
eu/fbk/dkm/pikes/resources/trec/FT.java | pikes-resources | 118 |
String docno = ""; if (docnoElement != null) { docno = docnoElement.getTextContent().trim(); } String date = ""; if (dateElement != null) { date = dateElement.getTextContent().trim(); } String headline = ""; if (headlineElement != null) { headline = headlineElement.getTextContent().trim(); } if (docno.equals("")) { LOGGER.error("DOCNO is empty"); } String url = String.format(urlTemplate, docno); headline = headline.replace('\n', ' '); headline = headline.replaceAll("\\s+", " "); text = text.replace('\n', ' '); text = text.replaceAll("\\s+", " "); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ecb/ECBevaluator2.java | pikes-resources | 76 |
eu/fbk/dkm/pikes/resources/ecb/ECBextractor.java | pikes-resources | 43 |
Set<String> okEvents = new HashSet<>(); for (final File file : Files.fileTreeTraverser().preOrderTraversal(inputNaf)) { if (!file.isFile()) { continue; } if (file.getName().startsWith(".")) { continue; } String path = file.getParentFile().toString(); Integer folder = Integer.parseInt(path.substring(path.lastIndexOf("/")).substring(1)); Integer fileNum = Integer.parseInt(file.getName().substring(0, file.getName().length() - 4)); LOGGER.debug(file.getAbsolutePath()); KAFDocument document = KAFDocument.createFromFile(file); |
File | Project | Line |
---|---|---|
ixa/kaflib/Predicate.java | pikes-naflib | 69 |
ixa/kaflib/Predicate.java | pikes-naflib | 216 |
public String getStr() { String str = ""; for (Term term : this.span.getTargets()) { if (!str.isEmpty()) { str += " "; } str += term.getStr(); } return str; } public ExternalRef getExternalRef(String resource) { for (ExternalRef ref : externalReferences) { if (ref.getResource().equalsIgnoreCase(resource)) { return ref; } } return null; } public List<ExternalRef> getExternalRefs() { return externalReferences; } public void addExternalRef(ExternalRef externalRef) { externalReferences.add(externalRef); } public void addExternalRefs(List<ExternalRef> externalRefs) { externalReferences.addAll(externalRefs); } public List<String> getFlags() { |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ecb/ECBevaluator.java | pikes-resources | 286 |
eu/fbk/dkm/pikes/resources/ecb/ECBevaluator2.java | pikes-resources | 78 |
eu/fbk/dkm/pikes/resources/ecb/ECBextractor.java | pikes-resources | 45 |
for (final File file : Files.fileTreeTraverser().preOrderTraversal(inputNaf)) { if (!file.isFile()) { continue; } if (file.getName().startsWith(".")) { continue; } String path = file.getParentFile().toString(); Integer folder = Integer.parseInt(path.substring(path.lastIndexOf("/")).substring(1)); Integer fileNum = Integer.parseInt(file.getName().substring(0, file.getName().length() - 4)); LOGGER.debug(file.getAbsolutePath()); KAFDocument document = KAFDocument.createFromFile(file); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/raid/pipeline/LinkLabeller.java | pikes-raid | 494 |
eu/fbk/dkm/pikes/raid/pipeline/SpanLabeller.java | pikes-raid | 513 |
} final Classifier classifier = Classifier.train(grid, this.trainingSet, ConfusionMatrix.labelComparator(PrecisionRecall.Measure.F1, 1, true), 100000); // Log parameters of the best classifier LOGGER.info("Best classifier parameters: {}", classifier.getParameters()); // Perform cross-validation and emit some performance statistics, if enabled if (analyze && LOGGER.isInfoEnabled()) { final List<LabelledVector> trainingPredictions = classifier.predict(false, this.trainingSet); final ConfusionMatrix matrix = LabelledVector.evaluate(this.trainingSet, trainingPredictions, 2); LOGGER.info("Performances on training set:\n{}", matrix); final ConfusionMatrix crossMatrix = Classifier.crossValidate( classifier.getParameters(), this.trainingSet, 5, -1); LOGGER.info("5-fold cross-validation performances:\n{}", crossMatrix); } // Build and return the created link labeller return new LinkLabeller(classifier, this.posPrefixes); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/naflib/Corpus.java | pikes-naflib | 109 |
eu/fbk/dkm/pikes/naflib/Corpus.java | pikes-naflib | 131 |
public Path file(final Object key) { try { int index; if (key instanceof Number) { index = ((Number) key).intValue(); } else if (key instanceof File) { index = Arrays.binarySearch(this.files, ((File) key).toPath()); } else if (key instanceof Path) { index = Arrays.binarySearch(this.files, key); } else { index = Arrays.binarySearch(this.files, Paths.get(key.toString())); } if (index < 0 || index >= this.files.length) { throw new IllegalArgumentException("No file in this corpus for " + key); } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/rdf/RDFGenerator.java | pikes-rdf | 745 |
eu/fbk/dkm/pikes/rdf/naf/NAFExtractor.java | pikes-rdf | 189 |
emitMeta(docURI, DCTERMS.LANGUAGE, ModelUtil.languageCodeToURI(this.document.getLang())); } // Emit an hash of the whitespace-normalized raw text, if available if (this.document.getRawText() != null) { final String rawText = this.document.getRawText(); final StringBuilder builder = new StringBuilder(); boolean addSpace = false; for (int i = 0; i < rawText.length(); ++i) { final char c = rawText.charAt(i); if (Character.isWhitespace(c)) { addSpace = builder.length() > 0; } else { if (addSpace) { builder.append(' '); addSpace = false; } builder.append(c); } } |
File | Project | Line |
---|---|---|
eu/fbk/shell/mdfsa/data/structures/TreeGraphStatisticalModel.java | pikes-raid | 215 |
eu/fbk/shell/mdfsa/data/structures/TreeGraphStatisticalModel.java | pikes-raid | 257 |
String key = new String(d[0] + "^^^" + d[2] + "^^^" + d[1]); t = domainModel.get(key); tI = domainIndependentModel.get(key); if(t == null) { t = new Tuple(1); t.setToken(key); t.setFrequency(1.0); } else { t.setFrequency(t.getFrequency() + 1.0); } if(tI == null) { tI = new Tuple(1); tI.setToken(key); tI.setFrequency(1.0); } else { tI.setFrequency(t.getFrequency() + 1.0); } domainModel.put(key, t); domainIndependentModel.put(key, tI); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ecb/ECBevaluator.java | pikes-resources | 117 |
eu/fbk/dkm/pikes/resources/ecb/ECBevaluator2.java | pikes-resources | 75 |
HashMap<String, String> goldClusters = new HashMap<>(); Set<String> okEvents = new HashSet<>(); for (final File file : Files.fileTreeTraverser().preOrderTraversal(inputNaf)) { if (!file.isFile()) { continue; } if (file.getName().startsWith(".")) { continue; } String path = file.getParentFile().toString(); Integer folder = Integer.parseInt(path.substring(path.lastIndexOf("/")).substring(1)); Integer fileNum = Integer.parseInt(file.getName().substring(0, file.getName().length() - 4)); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ecb/ECBmerger.java | pikes-resources | 47 |
eu/fbk/dkm/pikes/resources/ecb/ECBparser.java | pikes-resources | 43 |
File outputFolder = cmd.getOptionValue("output", File.class); if (!outputFolder.exists()) { outputFolder.mkdirs(); } DocumentBuilderFactory dbFactory = DocumentBuilderFactory.newInstance(); DocumentBuilder dBuilder = dbFactory.newDocumentBuilder(); // uncomment to get the manual mention spans //Pattern MY_PATTERN = Pattern.compile("\\\">[^<]*</MENTION>"); String tags; int i = 0; for (final File file : Files.fileTreeTraverser().preOrderTraversal(inputFolder)) { if (!file.isFile()) { continue; } if (file.getName().startsWith(".")) { continue; } String path = file.getParentFile().toString(); String folder = path.substring(path.lastIndexOf("/")); String local_name = folder + File.separator + file.getName(); String naf = nafFolder + File.separator + folder + File.separator + file.getName() |
File | Project | Line |
---|---|---|
eu/fbk/shell/mdfsa/data/structures/TreeGraphStatisticalModel.java | pikes-raid | 215 |
eu/fbk/shell/mdfsa/data/structures/TreeGraphStatisticalModel.java | pikes-raid | 236 |
String key = new String(d[0] + "^^^" + d[2] + "^^^" + d[1]); t = domainModel.get(key); tI = domainIndependentModel.get(key); if(t == null) { t = new Tuple(1); t.setToken(key); t.setFrequency(1.0); } else { t.setFrequency(t.getFrequency() + 1.0); } if(tI == null) { tI = new Tuple(1); tI.setToken(key); tI.setFrequency(1.0); } else { tI.setFrequency(t.getFrequency() + 1.0); } domainModel.put(key, t); domainIndependentModel.put(key, tI); /* Key: governor-dependent */ key = new String(d[1] + "^^^" + d[2]); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/tintopclient/TintopOrchestrator.java | pikes-tintop-client | 303 |
eu/fbk/dkm/pikes/tintop/FolderOrchestrator.java | pikes-tintop | 289 |
HashSet<String> skipPatterns = new HashSet<>(); if (skip != null) { BufferedReader reader = new BufferedReader(new FileReader(skip)); String line; while ((line = reader.readLine()) != null) { line = line.trim(); skipPatterns.add(line); } reader.close(); } if (!input.exists()) { logger.error("Input folder does not exist"); System.exit(1); } if (!output.exists()) { if (!output.mkdirs()) { logger.error("Unable to create output folder"); System.exit(1); } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/rdf/Renderer.java | pikes-rdf | 79 |
eu/fbk/dkm/pikes/rdf/naf/NAFRenderer.java | pikes-rdf | 53 |
public static final Map<Object, String> DEFAULT_COLOR_MAP = ImmutableMap .<Object, String>builder() // .put("node", "#F0F0F0") // .put(NWR.PERSON, "#FFC8C8") // .put(NWR.ORGANIZATION, "#FFFF84") // .put(NWR.LOCATION, "#A9C5EB") // .put(KS.ATTRIBUTE, "#EEBBEE") // // .put(KS.MONEY, "#EEBBEE") // // .put(KS.FACILITY, "#FFC65B") // // .put(KS.PRODUCT, "#FFC65B") // // .put(KS.WORK_OF_ART, "#FFC65B") // .put(SUMO.PROCESS, "#CFE990") // .put(SUMO.RELATION, "#FFFFFF") // .put(OWLTIME.INTERVAL, "#B4D1B6") // .put(OWLTIME.DATE_TIME_INTERVAL, "#B4D1B6") // .put(OWLTIME.PROPER_INTERVAL, "#B4D1B6") // .put(NWR.MISC, "#D1BAA2") // // .put(KS.LAW, "#D1BAA2") // .build(); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/rdf/Renderer.java | pikes-rdf | 400 |
eu/fbk/dkm/pikes/rdf/naf/NAFRenderer.java | pikes-rdf | 337 |
final Model mentionModel = new LinkedHashModel(); for (final Statement statement : model.filter(mentionID, null, null)) { final URI pred = statement.getPredicate(); if (!NIF.BEGIN_INDEX.equals(pred) && !NIF.END_INDEX.equals(pred) && !NIF.ANCHOR_OF.equals(pred) && !RDF.TYPE.equals(pred) && !KS.MENTION_OF.equals(pred)) { mentionModel.add(statement); } } if (!mentionModel.isEmpty()) { renderProperties(out, mentionModel, mentionID, false); } out.append("</td><td>"); renderObject(out, model.filter(null, GAF.DENOTED_BY, mentionID).subjects(), model); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/trec/CR.java | pikes-resources | 91 |
eu/fbk/dkm/pikes/resources/trec/FR94.java | pikes-resources | 150 |
eu/fbk/dkm/pikes/resources/trec/FT.java | pikes-resources | 88 |
eu/fbk/dkm/pikes/resources/trec/LATIMES.java | pikes-resources | 95 |
StringBuffer stringBuffer = new StringBuffer(); stringBuffer.append("<ROOT>\n"); stringBuffer.append(Files.toString(inputFile, Charsets.UTF_8)); stringBuffer.append("\n</ROOT>\n"); InputStream is = new ByteArrayInputStream(stringBuffer.toString().getBytes()); DocumentBuilderFactory dbFactory = DocumentBuilderFactory.newInstance(); DocumentBuilder dBuilder = dbFactory.newDocumentBuilder(); Document doc = dBuilder.parse(is); doc.getDocumentElement().normalize(); int i = 0; for (Element element : JOOX.$(doc).find("DOC")) { Element docnoElement = JOOX.$(element).find("DOCNO").get(0); Element dateElement = JOOX.$(element).find("DATE").get(0); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/trec/CR.java | pikes-resources | 115 |
eu/fbk/dkm/pikes/resources/trec/FBIS.java | pikes-resources | 175 |
LOGGER.error("TEXT is null"); continue; } String docno = ""; if (docnoElement != null) { docno = docnoElement.getTextContent().trim(); } String date = ""; if (dateElement != null) { date = dateElement.getTextContent().trim(); } String headline = ""; if (headlineElement != null) { headline = headlineElement.getTextContent().trim(); } if (docno.equals("")) { LOGGER.error("DOCNO is empty"); } String url = String.format(urlTemplate, docno); headline = headline.replace('\n', ' '); headline = headline.replaceAll("\\s+", " "); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/mpqa/CorpusPreprocessor.java | pikes-resources | 379 |
eu/fbk/dkm/pikes/resources/mpqa/CorpusPreprocessor.java | pikes-resources | 422 |
} // Holder String sources = thisRecord.getValue("nested-source"); addSources(sources, agentRecords, lastRecords, sentenceSpan, fields, documentURI, text); // Emit TSV string tsvWriter.append(getTsvString(documentURI, fields)); // Print debug LOGGER.debug(fields.get("type").toString()); LOGGER.debug(fields.toString()); for (String expression : fields.get("expression")) { LOGGER.debug(expression); Span span = new Span(expression); LOGGER.debug(span.apply(text)); } System.out.println(); } // Expressive for (final Record thisRecord : annotations.getRecords("GATE_expressive-subjectivity")) { |
File | Project | Line |
---|---|---|
eu/fbk/shell/mdfsa/data/structures/TreeGraphStatisticalModel.java | pikes-raid | 236 |
eu/fbk/shell/mdfsa/data/structures/TreeGraphStatisticalModel.java | pikes-raid | 257 |
key = new String(d[1] + "^^^" + d[2]); t = domainModel.get(key); tI = domainIndependentModel.get(key); if(t == null) { t = new Tuple(1); t.setToken(key); t.setFrequency(1.0); } else { t.setFrequency(t.getFrequency() + 1.0); } if(tI == null) { tI = new Tuple(1); tI.setToken(key); tI.setFrequency(1.0); } else { tI.setFrequency(t.getFrequency() + 1.0); } domainModel.put(key, t); domainIndependentModel.put(key, tI); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/tintop/annotators/raw/UKB.java | pikes-tintop | 61 |
eu/fbk/dkm/pikes/tintop/annotators/raw/UKB_MT.java | pikes-tintop | 48 |
String line = brCleanUp.readLine(); } private static void addTokenToContext(HashMap<String, String> term, char pos, int index, StringBuffer sb, HashMap<String, HashMap<String, String>> backupTerms) { String thisID = "w" + index; sb.append(term.get("lemma").toLowerCase().replace(' ', '-').replace('#', '.')); sb.append("#"); sb.append(pos); sb.append("#"); sb.append(thisID); sb.append("#1"); sb.append(" "); backupTerms.put(thisID, term); } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/trec/CR.java | pikes-resources | 119 |
eu/fbk/dkm/pikes/resources/trec/FT.java | pikes-resources | 118 |
String docno = ""; if (docnoElement != null) { docno = docnoElement.getTextContent().trim(); } String date = ""; if (dateElement != null) { date = dateElement.getTextContent().trim(); } String headline = ""; if (headlineElement != null) { headline = headlineElement.getTextContent().trim(); } if (docno.equals("")) { LOGGER.error("DOCNO is empty"); } String url = String.format(urlTemplate, docno); headline = headline.replace('\n', ' '); headline = headline.replaceAll("\\s+", " "); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/tintop/util/framenet/MergeMateFramenet.java | pikes-tintop | 1354 |
eu/fbk/dkm/pikes/tintop/util/framenet/MergeMateFramenet.java | pikes-tintop | 1390 |
String value = outputMappingsForRoles.get(outputMapping).get(key); String[] parts = value.split("@"); if (parts.length < 2) { LOGGER.error("This is impossible!"); break; } URI fbURI = FrameBase.propertyFor(parts[0], parts[1]); if (fbURI == null) { LOGGER.error("This should never happen!"); LOGGER.debug(key); LOGGER.debug(value); break; } writer.append(outputMapping.toString()).append('\t'); writer.append(key).append('\t'); writer.append(value).append('\t'); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/rdf/Renderer.java | pikes-rdf | 604 |
eu/fbk/dkm/pikes/rdf/naf/NAFRenderer.java | pikes-rdf | 571 |
return this; } public Builder withColorMap(@Nullable final Map<Object, String> colorMap) { this.colorMap = colorMap; return this; } public Builder withStyleMap(@Nullable final Map<Object, String> styleMap) { this.styleMap = styleMap; return this; } public Builder withTemplate(@Nullable final Object template) { this.template = template == null ? null : loadTemplate(template); return this; } public Builder withTemplateParameter(final String name, @Nullable final Object value) { this.templateParameters.put(name, value); return this; } public Renderer build() { return new Renderer(this); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ontonotes/frames/Vnrole.java | pikes-resources | 20 |
eu/fbk/dkm/pikes/resources/util/propbank/Vnrole.java | pikes-resources | 18 |
@XmlAccessorType(XmlAccessType.FIELD) @XmlType(name = "") @XmlRootElement(name = "vnrole") public class Vnrole { @XmlAttribute(name = "vncls", required = true) @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String vncls; @XmlAttribute(name = "vntheta", required = true) @XmlJavaTypeAdapter(CollapsedStringAdapter.class) protected String vntheta; /** * Recupera il valore della proprietà vncls. * * @return * possible object is * {@link String } * */ public String getVncls() { return vncls; } /** * Imposta il valore della proprietà vncls. * * @param value * allowed object is * {@link String } * */ public void setVncls(String value) { this.vncls = value; } /** * Recupera il valore della proprietà vntheta. * * @return * possible object is * {@link String } * */ public String getVntheta() { return vntheta; } /** * Imposta il valore della proprietà vntheta. * * @param value * allowed object is * {@link String } * */ public void setVntheta(String value) { this.vntheta = value; } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ecb/ECBevaluator.java | pikes-resources | 118 |
eu/fbk/dkm/pikes/resources/ecb/ECBextractor.java | pikes-resources | 43 |
Set<String> okEvents = new HashSet<>(); for (final File file : Files.fileTreeTraverser().preOrderTraversal(inputNaf)) { if (!file.isFile()) { continue; } if (file.getName().startsWith(".")) { continue; } String path = file.getParentFile().toString(); Integer folder = Integer.parseInt(path.substring(path.lastIndexOf("/")).substring(1)); Integer fileNum = Integer.parseInt(file.getName().substring(0, file.getName().length() - 4)); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ecb/ECBevaluator.java | pikes-resources | 120 |
eu/fbk/dkm/pikes/resources/ecb/ECBevaluator.java | pikes-resources | 286 |
for (final File file : Files.fileTreeTraverser().preOrderTraversal(inputNaf)) { if (!file.isFile()) { continue; } if (file.getName().startsWith(".")) { continue; } String path = file.getParentFile().toString(); Integer folder = Integer.parseInt(path.substring(path.lastIndexOf("/")).substring(1)); Integer fileNum = Integer.parseInt(file.getName().substring(0, file.getName().length() - 4)); |
File | Project | Line |
---|---|---|
eu/fbk/shell/mdfsa/data/structures/DomainGraph.java | pikes-raid | 74 |
eu/fbk/shell/mdfsa/data/structures/DomainGraph.java | pikes-raid | 180 |
int polarity = di.getPolarity(); for(String currentFeature: features) { ArrayList<Long> featureIds = this.graph.getFeatureIds(currentFeature); if(featureIds != null) { for(Long featureId: featureIds) { Double currentPolarity = this.startPolarities.get(featureId); if(currentPolarity == null) { currentPolarity = new Double(polarity); this.startPolarities.put(featureId, currentPolarity); this.currentPolarities.put(featureId, currentPolarity); } else { currentPolarity += polarity; this.startPolarities.put(featureId, currentPolarity); this.currentPolarities.put(featureId, currentPolarity); } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/trec/CR.java | pikes-resources | 93 |
eu/fbk/dkm/pikes/resources/trec/FBIS.java | pikes-resources | 146 |
eu/fbk/dkm/pikes/resources/trec/FT.java | pikes-resources | 90 |
eu/fbk/dkm/pikes/resources/trec/LATIMES.java | pikes-resources | 97 |
stringBuffer.append(Files.toString(inputFile, Charsets.UTF_8)); stringBuffer.append("\n</ROOT>\n"); InputStream is = new ByteArrayInputStream(stringBuffer.toString().getBytes()); DocumentBuilderFactory dbFactory = DocumentBuilderFactory.newInstance(); DocumentBuilder dBuilder = dbFactory.newDocumentBuilder(); Document doc = dBuilder.parse(is); doc.getDocumentElement().normalize(); int i = 0; for (Element element : JOOX.$(doc).find("DOC")) { Element docnoElement = JOOX.$(element).find("DOCNO").get(0); Element dateElement = JOOX.$(element).find("DATE").get(0); |
File | Project | Line |
---|---|---|
ixa/kaflib/ReadWriteManager.java | pikes-naflib | 535 |
ixa/kaflib/ReadWriteManager.java | pikes-naflib | 578 |
"Every property must contain a 'span' element inside 'references'"); } List<Span<Term>> references = new ArrayList<Span<Term>>(); for (Element spanElem : spanElems) { Span<Term> span = kaf.newTermSpan(); List<Element> targetElems = spanElem.getChildren(); if (targetElems.size() < 1) { throw new IllegalStateException( "Every span in a property must contain at least one target inside"); } for (Element targetElem : targetElems) { String targetTermId = getAttribute("id", targetElem); Term targetTerm = termIndex.get(targetTermId); if (targetTerm == null) { throw new KAFNotValidException( "Term " + targetTermId + " not found when loading property " + pid); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ontonotes/senses/WORDMETA.java | pikes-resources | 19 |
eu/fbk/dkm/pikes/resources/util/onsenses/WORDMETA.java | pikes-resources | 23 |
@XmlAccessorType(XmlAccessType.FIELD) @XmlType(name = "") @XmlRootElement(name = "WORD_META") public class WORDMETA { @XmlAttribute(name = "authors", required = true) @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String authors; @XmlAttribute(name = "sample_score") @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String sampleScore; /** * Recupera il valore della proprietà authors. * * @return * possible object is * {@link String } * */ public String getAuthors() { return authors; } /** * Imposta il valore della proprietà authors. * * @param value * allowed object is * {@link String } * */ public void setAuthors(String value) { this.authors = value; } /** * Recupera il valore della proprietà sampleScore. * * @return * possible object is * {@link String } * */ public String getSampleScore() { return sampleScore; } /** * Imposta il valore della proprietà sampleScore. * * @param value * allowed object is * {@link String } * */ public void setSampleScore(String value) { this.sampleScore = value; } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/tintopclient/TintopOrchestrator.java | pikes-tintop-client | 67 |
eu/fbk/dkm/pikes/tintop/FolderOrchestrator.java | pikes-tintop | 86 |
while (true) { String filename = null; try { filename = getNextFile(session); if (filename == null) { break; } File file = new File(filename); if (!file.exists()) { break; } File outputFile = getOutputFile(file, session); // todo: use parameters outputFile = new File(outputFile.getAbsolutePath() + ".gz"); logger.debug("Output file: " + outputFile); logger.info("Loading file: " + filename); BufferedReader reader = new BufferedReader(new FileReader(filename)); String whole = IOUtils.toString(reader); reader.close(); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ecb/ECBPlusEvaluator.java | pikes-resources | 39 |
eu/fbk/dkm/pikes/resources/ecb/ECBPlusEvaluatorLemma.java | pikes-resources | 40 |
eu/fbk/dkm/pikes/resources/ecb/ECBevaluator.java | pikes-resources | 41 |
writer.append(token.getMorphofeat()).append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ecb/ECBPlusEvaluator.java | pikes-resources | 40 |
eu/fbk/dkm/pikes/resources/ecb/ECBPlusEvaluatorLemma.java | pikes-resources | 39 |
eu/fbk/dkm/pikes/resources/ecb/ECBevaluator.java | pikes-resources | 40 |
writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append("_").append("\t"); writer.append(last); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/mpqa/CorpusPreprocessor.java | pikes-resources | 379 |
eu/fbk/dkm/pikes/resources/mpqa/CorpusPreprocessor.java | pikes-resources | 466 |
} // Holder String sources = thisRecord.getValue("nested-source"); addSources(sources, agentRecords, lastRecords, sentenceSpan, fields, documentURI, text); // Emit TSV string tsvWriter.append(getTsvString(documentURI, fields)); // Print debug LOGGER.debug(fields.get("type").toString()); LOGGER.debug(fields.toString()); for (String expression : fields.get("expression")) { LOGGER.debug(expression); Span span = new Span(expression); LOGGER.debug(span.apply(text)); } System.out.println(); } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/raid/CreateTrainingForExpression.java | pikes-raid | 559 |
eu/fbk/dkm/pikes/raid/UpdateNafsWithResults.java | pikes-raid | 109 |
SentiWordNet.init(); } LOGGER.info("Parsing corpus"); Corpus[] corpuses = new Corpus[2]; if (trainList != null) { List<File> trainFiles = readList(trainList, mainFolder, "naf"); List<File> testFiles = readList(testList, mainFolder, "naf"); corpuses[0] = Corpus.create(false, trainFiles); corpuses[1] = Corpus.create(false, testFiles); } else { Corpus myCorpus = Corpus.create(false, mainFolder); corpuses = myCorpus.split(seed, split, 1.0f - split); } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ecb/ECBPlusEvaluator.java | pikes-resources | 244 |
eu/fbk/dkm/pikes/resources/ecb/ECBPlusEvaluatorLemma.java | pikes-resources | 244 |
eu/fbk/dkm/pikes/resources/ecb/ECBevaluator.java | pikes-resources | 189 |
theBaseline.put(id, last); } printToken(baselineBuilder, token, n, last); } goldBuilder.append("\n"); baselineBuilder.append("\n"); if (useThis) { goldWriter.append(goldBuilder.toString()); baselineWriter.append(baselineBuilder.toString()); } } // break; } goldWriter.close(); baselineWriter.close(); // Set<Set> goldClusters = new HashSet<>(); // for (String key : goldTmpClusters.keySet()) { // Set<String> cluster = goldTmpClusters.get(key); // if (cluster.size() > 1 || !removeAloneClusters) { // goldClusters.add(cluster); // } // } // LOGGER.info("Gold clusters: {}", goldClusters.size()); in = new FileReader(inputCsv); records = CSVFormat.EXCEL.withHeader().parse(in); // Size must be always 4! int clusterID = 0; HashMap<String, Integer> clusterIndexes = new HashMap<>(); HashMultimap<Integer, String> theClusters = HashMultimap.create(); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/rdf/Renderer.java | pikes-rdf | 348 |
eu/fbk/dkm/pikes/rdf/naf/NAFRenderer.java | pikes-rdf | 283 |
out.append(shorten(KS.EXPRESSED_BY)); out.append("</th></tr>\n"); out.append("</thead>\n<tbody>\n"); for (final Statement statement : this.statementComparator.sortedCopy(model)) { if (statement.getContext() != null) { out.append("<tr><td>"); renderObject(out, statement.getSubject(), model); out.append("</td><td>"); renderObject(out, statement.getPredicate(), model); out.append("</td><td>"); renderObject(out, statement.getObject(), model); out.append("</td><td>"); String separator = ""; for (final Value mentionID : model.filter(statement.getContext(), KS.EXPRESSED_BY, |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/trec/FBIS.java | pikes-resources | 75 |
eu/fbk/dkm/pikes/resources/trec/FR94.java | pikes-resources | 72 |
eu/fbk/dkm/pikes/resources/trec/FT.java | pikes-resources | 67 |
if (file.getName().startsWith(".")) { continue; } String outputTemplate = outputDir.getAbsolutePath() + File.separator + file.getName(); File newFolder = new File(outputTemplate); newFolder.mkdirs(); outputTemplate += File.separator + "NAF"; saveFile(file, outputTemplate, urlTemplate); } } catch (Exception e) { CommandLine.fail(e); } } private static void saveFile(File inputFile, String outputFilePattern, String urlTemplate) throws IOException, SAXException, ParserConfigurationException { LOGGER.info("Input file: {}", inputFile); StringBuffer stringBuffer = new StringBuffer(); stringBuffer.append("<?xml version=\"1.0\"?>\n" |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/vuaopinion/CorpusAnnotator.java | pikes-resources | 37 |
eu/fbk/dkm/pikes/resources/vuaopinion/CorpusPreprocessor.java | pikes-resources | 61 |
boolean forceOpinion = cmd.hasOption("f"); final File inputPath = cmd.getOptionValue("i", File.class); if (!inputPath.exists()) { throw new IOException(String.format("Folder %s does not exist", inputPath.getAbsolutePath())); } File kafPath = new File(inputPath.getAbsolutePath() + File.separator + "kaf"); if (!kafPath.exists()) { throw new IOException(String.format("Folder %s does not exist", kafPath.getAbsolutePath())); } File nafPath = new File(inputPath.getAbsolutePath() + File.separator + "naf-parsed"); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/tintop/annotators/AnnaParseAnnotator.java | pikes-tintop | 60 |
eu/fbk/dkm/pikes/tintop/util/framenet/ParseFrameNetExamplesWithMate.java | pikes-tintop | 194 |
String lemma = token.get(CoreAnnotations.LemmaAnnotation.class); form = AnnotatorUtils.codeToParenthesis(form); lemma = AnnotatorUtils.codeToParenthesis(lemma); pos = AnnotatorUtils.codeToParenthesis(pos); forms.add(form); poss.add(pos); lemmas.add(lemma); } SentenceData09 localSentenceData091 = new SentenceData09(); localSentenceData091.init(forms.toArray(new String[forms.size()])); localSentenceData091.setPPos(poss.toArray(new String[poss.size()])); SentenceData09 localSentenceData092; |
File | Project | Line |
---|---|---|
ixa/kaflib/Coref.java | pikes-naflib | 51 |
ixa/kaflib/Entity.java | pikes-naflib | 56 |
ixa/kaflib/Feature.java | pikes-naflib | 41 |
for (Span<Term> span : coref.getSpans()) { /* Copy span */ List<Term> targets = span.getTargets(); List<Term> copiedTargets = new ArrayList<Term>(); for (Term term : targets) { Term copiedTerm = terms.get(term.getId()); if (copiedTerm == null) { throw new IllegalStateException("Term not found when copying " + id); } copiedTargets.add(copiedTerm); } if (span.hasHead()) { Term copiedHead = terms.get(span.getHead().getId()); this.mentions.add(new Span<Term>(copiedTargets, copiedHead)); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ontonotes/frames/Frameset.java | pikes-resources | 19 |
eu/fbk/dkm/pikes/resources/util/propbank/Frameset.java | pikes-resources | 17 |
@XmlAccessorType(XmlAccessType.FIELD) @XmlType(name = "", propOrder = { "noteOrPredicate" }) @XmlRootElement(name = "frameset") public class Frameset { @XmlElements({ @XmlElement(name = "note", type = Note.class), @XmlElement(name = "predicate", type = Predicate.class) }) protected List<Object> noteOrPredicate; /** * Gets the value of the noteOrPredicate property. * * <p> * This accessor method returns a reference to the live list, * not a snapshot. Therefore any modification you make to the * returned list will be present inside the JAXB object. * This is why there is not a <CODE>set</CODE> method for the noteOrPredicate property. * * <p> * For example, to add a new item, do as follows: * <pre> * getNoteOrPredicate().add(newItem); * </pre> * * * <p> * Objects of the following type(s) are allowed in the list * {@link Note } * {@link Predicate } * * */ public List<Object> getNoteOrPredicate() { if (noteOrPredicate == null) { noteOrPredicate = new ArrayList<Object>(); } return this.noteOrPredicate; } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ontonotes/frames/Roles.java | pikes-resources | 19 |
eu/fbk/dkm/pikes/resources/util/propbank/Roles.java | pikes-resources | 17 |
@XmlAccessorType(XmlAccessType.FIELD) @XmlType(name = "", propOrder = { "noteOrRole" }) @XmlRootElement(name = "roles") public class Roles { @XmlElements({ @XmlElement(name = "note", type = Note.class), @XmlElement(name = "role", type = Role.class) }) protected List<Object> noteOrRole; /** * Gets the value of the noteOrRole property. * * <p> * This accessor method returns a reference to the live list, * not a snapshot. Therefore any modification you make to the * returned list will be present inside the JAXB object. * This is why there is not a <CODE>set</CODE> method for the noteOrRole property. * * <p> * For example, to add a new item, do as follows: * <pre> * getNoteOrRole().add(newItem); * </pre> * * * <p> * Objects of the following type(s) are allowed in the list * {@link Note } * {@link Role } * * */ public List<Object> getNoteOrRole() { if (noteOrRole == null) { noteOrRole = new ArrayList<Object>(); } return this.noteOrRole; } } |
File | Project | Line |
---|---|---|
ixa/kaflib/Entity.java | pikes-naflib | 109 |
ixa/kaflib/Feature.java | pikes-naflib | 89 |
} /** Returns the term targets of the first span. When targets of other spans are needed getReferences() method should be used. */ public List<Term> getTerms() { return this.references.get(0).getTargets(); } /** Adds a term to the first span. */ public void addTerm(Term term) { this.references.get(0).addTarget(term); } /** Adds a term to the first span. */ public void addTerm(Term term, boolean isHead) { this.references.get(0).addTarget(term, isHead); } public List<Span<Term>> getSpans() { return this.references; } public void addSpan(Span<Term> span) { |
File | Project | Line |
---|---|---|
ixa/kaflib/ReadWriteManager.java | pikes-naflib | 334 |
ixa/kaflib/ReadWriteManager.java | pikes-naflib | 379 |
List<Span<Term>> references = new ArrayList<Span<Term>>(); for (Element spanElem : spanElems) { Span<Term> span = kaf.newTermSpan(); List<Element> targetElems = spanElem.getChildren(); if (targetElems.size() < 1) { throw new IllegalStateException( "Every span in an entity must contain at least one target inside"); } for (Element targetElem : targetElems) { String targetTermId = getAttribute("id", targetElem); Term targetTerm = termIndex.get(targetTermId); if (targetTerm == null) { throw new KAFNotValidException( "Term " + targetTermId + " not found when loading entity " + entId); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/mpqa/JohanssonAnnotator.java | pikes-resources | 185 |
eu/fbk/dkm/pikes/resources/mpqa/JohanssonAnnotator.java | pikes-resources | 272 |
eu/fbk/dkm/pikes/resources/mpqa/JohanssonAnnotator.java | pikes-resources | 333 |
for (LKAnnotationEntity entity : dseLayer.entityList) { Span<Term> expressionSpan = getSpanFromEntity(entity, document); Span<Term> holderSpan = KAFDocument.newTermSpan(); // Span<Term> targetSpan = KAFDocument.newTermSpan(); if (expressionSpan.size() == 0) { LOGGER.debug("Expression span is empty [{}/{}]", nafFile.getName(), entity.localURI); continue; } Integer sentence = sentenceForSpan(expressionSpan); if (sentence == null) { LOGGER.warn("Expression span is not in sentence [{}/{}]", nafFile.getName(), entity.localURI); continue; } Opinion opinion = document.newOpinion(); opinion.setLabel("gold-eu.fbk.dkm.pikes.resources.mpqa-subjective"); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ontonotes/frames/Rel.java | pikes-resources | 19 |
eu/fbk/dkm/pikes/resources/util/propbank/Rel.java | pikes-resources | 17 |
@XmlAccessorType(XmlAccessType.FIELD) @XmlType(name = "", propOrder = { "value" }) @XmlRootElement(name = "rel") public class Rel { @XmlAttribute(name = "f") @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String f; @XmlValue protected String value; /** * Recupera il valore della proprietà f. * * @return * possible object is * {@link String } * */ public String getF() { return f; } /** * Imposta il valore della proprietà f. * * @param value * allowed object is * {@link String } * */ public void setF(String value) { this.f = value; } /** * Recupera il valore della proprietà value. * * @return * possible object is * {@link String } * */ public String getvalue() { return value; } /** * Imposta il valore della proprietà value. * * @param value * allowed object is * {@link String } * */ public void setvalue(String value) { this.value = value; } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/ontonotes/frames/Text.java | pikes-resources | 19 |
eu/fbk/dkm/pikes/resources/util/propbank/Text.java | pikes-resources | 17 |
@XmlAccessorType(XmlAccessType.FIELD) @XmlType(name = "", propOrder = { "value" }) @XmlRootElement(name = "text") public class Text { @XmlAttribute(name = "src") @XmlJavaTypeAdapter(NormalizedStringAdapter.class) protected String src; @XmlValue protected String value; /** * Recupera il valore della proprietà src. * * @return * possible object is * {@link String } * */ public String getSrc() { return src; } /** * Imposta il valore della proprietà src. * * @param value * allowed object is * {@link String } * */ public void setSrc(String value) { this.src = value; } /** * Recupera il valore della proprietà value. * * @return * possible object is * {@link String } * */ public String getvalue() { return value; } /** * Imposta il valore della proprietà value. * * @param value * allowed object is * {@link String } * */ public void setvalue(String value) { this.value = value; } } |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/resources/trec/FBIS.java | pikes-resources | 146 |
eu/fbk/dkm/pikes/resources/trec/FR94.java | pikes-resources | 152 |
.replaceAll("</?3>", "") ); stringBuffer.append("\n</ROOT>\n"); InputStream is = new ByteArrayInputStream(stringBuffer.toString().getBytes()); DocumentBuilderFactory dbFactory = DocumentBuilderFactory.newInstance(); DocumentBuilder dBuilder = dbFactory.newDocumentBuilder(); Document doc = dBuilder.parse(is); doc.getDocumentElement().normalize(); int i = 0; for (Element element : JOOX.$(doc).find("DOC")) { Element docnoElement = JOOX.$(element).find("DOCNO").get(0); Element dateElement = JOOX.$(element).find("DATE1").get(0); |
File | Project | Line |
---|---|---|
eu/fbk/dkm/pikes/tintopclient/TintopOrchestrator.java | pikes-tintop-client | 93 |
eu/fbk/dkm/pikes/tintop/FolderOrchestrator.java | pikes-tintop | 122 |
logger.debug(naf); if (naf != null) { logger.info("Writing file " + outputFile); Files.createParentDirs(outputFile); try (Writer w = IO.utf8Writer(IO.buffer(IO.write(outputFile.getAbsolutePath())))) { w.write(naf); } } } catch (final Throwable ex) { logger.error(filename + " --- " + ex.getMessage()); markFileAsNotDone(filename); try { logger.info("Sleeping..."); Thread.sleep(sleepingTime); |