Package opennlp.tools.util

Examples of opennlp.tools.util.Span

88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
        }

        List<LocationOccurrence> nerResults = new ArrayList<LocationOccurrence>();

        // The values used in these Spans are string character offsets
        Span sentenceSpans[] = sentenceDetector.sentPosDetect(plainText);

        // Each sentence gets processed on its own
        for (Span sentenceSpan : sentenceSpans) {

            // find the start and end position of this sentence in the document
            String sentence = plainText.substring(sentenceSpan.getStart(), sentenceSpan.getEnd());

            // tokenize the text into the required OpenNLP format
            String[] tokens = tokenizer.tokenize(sentence);

            //the values used in these Spans are string character offsets of each token from the sentence beginning
            Span[] tokenPositionsWithinSentence = tokenizer.tokenizePos(sentence);

            // find the location names in the tokenized text
            // the values used in these Spans are NOT string character offsets, they are indices into the 'tokens' array
            Span names[] = nameFinder.find(tokens);


            //for each name that got found, create our corresponding occurrence
            for (Span name : names) {
View Full Code Here
76
77
78
79
80
81
82
83
84
85
86
87
88
89
   
    if (tokenOffset >= sentences.length) {
      return false;
    }
   
    Span sentenceSpan = sentences[tokenOffset];
    clearAttributes();
    int start = sentenceSpan.getStart();
    int end   = sentenceSpan.getEnd();
    termAtt.copyBuffer(inputSentence, start, end-start);
    posIncrAtt.setPositionIncrement(1);
    offsetAtt.setOffset(start, end);
    tokenOffset++;
   
View Full Code Here
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
  */
  //<end id="ne-remove-conflicts"/>
  @Test
  public void testRemoveConflicts() {
    List<Annotation> annotations = new ArrayList<Annotation>();
    annotations.add(new Annotation("person", new Span(1, 5), 0.75));
    annotations.add(new Annotation("person", new Span(7, 10), 0.95));
    annotations.add(new Annotation("location", new Span(11, 15), 0.85));
    removeConflicts(annotations);
    assertTrue(annotations.size() == 3);
    annotations.add(new Annotation("location", new Span(2, 7), 0.85));
    removeConflicts(annotations);
    assertTrue(annotations.size() == 3);
    assertTrue(((Annotation) annotations.get(0)).getSpan().getStart() == 2);
    annotations.clear();
    annotations.add(new Annotation("person", new Span(1, 5), 0.75));
    annotations.add(new Annotation("person", new Span(7, 10), 0.95));
    annotations.add(new Annotation("location", new Span(11, 15), 0.85));
    annotations.add(new Annotation("person", new Span(3, 8), 0.85));
    removeConflicts(annotations);
    assertTrue(annotations.size() == 2);
    assertTrue(((Annotation) annotations.get(0)).getSpan().getStart() == 7);
  }
View Full Code Here
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
      Span[] tokenSpans = tokenizer.tokenizePos(sentences[si]); //<co id="co.opennlp.name.tokenizepos"/>
      String[] tokens = Span.spansToStrings(tokenSpans, sentences[si]); //<co id="co.opennlp.name.convert2strings"/>
      Span[] names = finder.find(tokens); //<co id="co.opennlp.name.findnames4"/>

      for (int ni = 0; ni < names.length; ni++) {
        Span startSpan = tokenSpans[names[ni].getStart()]; //<co id="co.opennlp.name.computestart"/>
        int nameStart  = startSpan.getStart();
       
        Span endSpan   = tokenSpans[names[ni].getEnd() - 1]; //<co id="co.opennlp.name.computeend"/>
        int nameEnd    = endSpan.getEnd();
       
        String name = sentences[si].substring(nameStart, nameEnd); //<co id="co.opennlp.name.namestring"/>
        System.out.println(name);
      }
    }
View Full Code Here
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
    int chunkStart = -1;
    String chunkType = null;
    double logProb=0;
    for (int ci=0,cn=chunks.length;ci<cn;ci++) {
      if (ci > 0 && !chunks[ci].startsWith("I-") && !chunks[ci-1].equals("O")) {
        Span span = new Span(children[chunkStart].getSpan().getStart(),children[ci-1].getSpan().getEnd());
        tokens.insert(new Parse(tokens.getText(), span, chunkType, logProb,children[ci-1]));
        logProb=0;
      }           
      if (chunks[ci].startsWith("B-")) {
        chunkStart = ci;
        chunkType = chunks[ci].substring(2);
      }
      logProb+=Math.log(probs[ci]);
    }
    if (!chunks[chunks.length-1].equals("O")) {
      int ci = chunks.length;
      Span span = new Span(children[chunkStart].getSpan().getStart(),children[ci-1].getSpan().getEnd());
      tokens.insert(new Parse(tokens.getText(), span, chunkType, logProb,children[ci-1]));
    }
    return tokens;
  }
View Full Code Here
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
  @Override
  protected Iterator<Event> createEvents(TokenSample tokenSample) {

    List<Event> events = new ArrayList<Event>(50);

    Span tokens[] = tokenSample.getTokenSpans();
    String text = tokenSample.getText();

    if (tokens.length > 0) {

      int start = tokens[0].getStart();
      int end = tokens[tokens.length - 1].getEnd();

      String sent = text.substring(start, end);

      Span[] candTokens = WhitespaceTokenizer.INSTANCE.tokenizePos(sent);

      int firstTrainingToken = -1;
      int lastTrainingToken = -1;
      for (Span candToken : candTokens) {
        Span cSpan = candToken;
        String ctok = sent.substring(cSpan.getStart(), cSpan.getEnd());
        //adjust cSpan to text offsets
        cSpan = new Span(cSpan.getStart() + start, cSpan.getEnd() + start);
        //should we skip this token
        if (ctok.length() > 1
          && (!skipAlphaNumerics || !alphaNumeric.matcher(ctok).matches())) {

          //find offsets of annotated tokens inside of candidate tokens
          boolean foundTrainingTokens = false;
          for (int ti = lastTrainingToken + 1; ti < tokens.length; ti++) {
            if (cSpan.contains(tokens[ti])) {
              if (!foundTrainingTokens) {
                firstTrainingToken = ti;
                foundTrainingTokens = true;
              }
              lastTrainingToken = ti;
            }
            else if (cSpan.getEnd() < tokens[ti].getEnd()) {
              break;
            }
            else if (tokens[ti].getEnd() < cSpan.getStart()) {
              //keep looking
            }
            else {
              if (logger.isLoggable(Level.WARNING)) {
                logger.warning("Bad training token: " + tokens[ti] + " cand: " + cSpan +
                    " token="+text.substring(tokens[ti].getStart(), tokens[ti].getEnd()));
              }
            }
          }

          // create training data
          if (foundTrainingTokens) {

            for (int ti = firstTrainingToken; ti <= lastTrainingToken; ti++) {
              Span tSpan = tokens[ti];
              int cStart = cSpan.getStart();
              for (int i = tSpan.getStart() + 1; i < tSpan.getEnd(); i++) {
                String[] context = cg.getContext(ctok, i - cStart);
                events.add(new Event(TokenizerME.NO_SPLIT, context));
              }

              if (tSpan.getEnd() != cSpan.getEnd()) {
                String[] context = cg.getContext(ctok, tSpan.getEnd() - cStart);
                events.add(new Event(TokenizerME.SPLIT, context));
              }
            }
          }
        }
View Full Code Here
120
121
122
123
124
125
126
127
      else {
        id = -1;
        // throw invalid format exception ...
      }
       
      mentionStack.push(new CorefMention(new Span(beginOffset, beginOffset), id, attributes.get("MIN")));
    }
  }
View Full Code Here
139
140
141
142
143
144
145
146
147
148
149
  @Override
  public void endElement(String name) {
   
    if (COREF_ELEMENT.equals(name)) {
      CorefMention mention = mentionStack.pop();
      mention.span = new Span(mention.span.getStart(), text.size());
      mentions.add(mention);
    }
   
    if (MucElementNames.CONTENT_ELEMENTS.contains(name)) {
     
View Full Code Here
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
  }
 
  static Parse createIncompleteParse(String tokens[]) {
   
    // produce text
    Span tokenSpans[] = new Span[tokens.length];
    StringBuilder textBuilder = new StringBuilder();
   
    for (int i = 0; i < tokens.length; i++) {
     
      if (textBuilder.length() > 0) {
        textBuilder.append(' ');
      }
     
      int startOffset = textBuilder.length();
      textBuilder.append(tokens[i]);
      tokenSpans[i] = new Span(startOffset, textBuilder.length());
    }
   
    String text = textBuilder.toString();
   
    Parse p = new Parse(text, new Span(0, text.length()), AbstractBottomUpParser.INC_NODE, 0, 0);
   
    for (int i = 0; i < tokenSpans.length; i++) {
      Span tokenSpan = tokenSpans[i];
      p.insert(new Parse(text, new Span(tokenSpan.getStart(), tokenSpan.getEnd()), AbstractBottomUpParser.TOK_NODE, 0, i));
    }
   
    return p;
  }
View Full Code Here
195
196
197
198
199
200
201
202
203
204
205
    List<Span> openNLPSpans = new LinkedList<Span>();

    while (containingTokens.hasNext()) {
      AnnotationFS tokenAnnotation = containingTokens.next();

      openNLPSpans.add(new Span(tokenAnnotation.getBegin()
          - sentence.getBegin(), tokenAnnotation.getEnd()
          - sentence.getBegin()));
    }

    Span[] spans = openNLPSpans.toArray(new Span[openNLPSpans.size()]);
View Full Code Here
TOP

Related Classes of opennlp.tools.util.Span

  • com.bericotech.clavin.extractor.ApacheExtractor
  • com.tamingtext.opennlp.NameFinderTest
  • net.sf.nlpshell.CorefParse
  • net.sf.nlpshell.Main
  • opennlp.tools.chunker.ChunkerMETest
  • opennlp.tools.chunker.ChunkSample
  • opennlp.tools.chunker.ChunkSampleTest
  • opennlp.tools.cmdline.EvaluationErrorPrinter
  • opennlp.tools.cmdline.namefind.TokenNameFinderTool
  • opennlp.tools.cmdline.parser.ParserTool
    • Copyright © 2018 www.massapicom. All rights reserved.
    All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.