Package opennlp.tools.parser

Examples of opennlp.tools.parser.Parse

44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
  public String[] getContext(Parse parent, Parse[] constituents, int index, boolean trimFrontier) {
    List<String> features = new ArrayList<String>(100);
    //default
    features.add("default");
    Parse[] children = Parser.collapsePunctuation(parent.getChildren(),punctSet);
    Parse pstart = children[0];
    Parse pend = children[children.length-1];
    String type = parent.getType();
    checkcons(pstart, "begin", type, features);
    checkcons(pend, "last", type, features);
    String production = "p="+production(parent,false);
    String punctProduction = "pp="+production(parent,true);
    features.add(production);
    features.add(punctProduction);


    Parse p1 = null;
    Parse p2 = null;
    Parse p_1 = null;
    Parse p_2 = null;
    Collection<Parse> p1s = constituents[index].getNextPunctuationSet();
    Collection<Parse> p2s = null;
    Collection<Parse> p_1s = constituents[index].getPreviousPunctuationSet();
    Collection<Parse> p_2s = null;
    List<Parse> rf;
View Full Code Here
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
   * @param constituents The constituents of the parse so far.
   * @param index The index of the constituent where a build decision is being made.
   * @return the contexts/features for the decision to build a new constituent.
   */
  public String[] getContext(Parse[] constituents, int index) {
    Parse p_2 = null;
    Parse p_1 = null;
    Parse p0 = null;
    Parse p1 = null;
    Parse p2 = null;
    int ps = constituents.length;

    p0 = constituents[index];
    if (index + 1 < ps) {
      p1 = constituents[index + 1];
View Full Code Here
58
59
60
61
62
63
64
65
66
67
68
    if (min != null) {
     
      int startOffset = p.toString().indexOf(min);
      int endOffset = startOffset + min.length();
     
      Parse tokens[] = p.getTagNodes();
     
      int beginToken = -1;
      int endToken = -1;
     
      for (int i = 0; i < tokens.length; i++) {
View Full Code Here
86
87
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
 
  public static boolean addMention(int id, Span mention, Parse[] tokens) {

  boolean failed = false;
   
    Parse startToken = tokens[mention.getStart()];
    Parse endToken = tokens[mention.getEnd() - 1];
    Parse commonParent = startToken.getCommonParent(endToken);
   
    if (commonParent != null) {
//      Span mentionSpan = new Span(startToken.getSpan().getStart(), endToken.getSpan().getEnd());
     
      if (entitySet.contains(commonParent.getType())) {
        commonParent.getParent().setType("NP#" + id);           
      }
      else if (commonParent.getType().equals("NML")) {
        commonParent.setType("NML#" + id);
      }
      else if (commonParent.getType().equals("NP")) {
        commonParent.setType("NP#" + id);
      }
      else {
        System.out.println("Inserting mention failed: " + commonParent.getType() + " Failed id: " + id);
        failed = true;
      }
    }
    else {
      throw new IllegalArgumentException("Tokens must always have a common parent!");
View Full Code Here
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
   * @param node The node whose parents are to be returned.
   * @return a set of parent nodes.
   */
  private Map<Parse, Integer> getNonAdjoinedParent(Parse node) {
    Map<Parse, Integer> parents = new HashMap<Parse, Integer>();
    Parse parent = node.getParent();
    int index = indexOf(node,parent);
    parents.put(parent, index);
    while(parent.getType().equals(node.getType())) {
      node = parent;
      parent = parent.getParent();
      index = indexOf(node,parent);
      parents.put(parent, index);
    }
    return parents;
  }
View Full Code Here
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
      List<CorefMention[]> allMentions = sample.getMentions();
      List<Parse> allParses = sample.getParses();
     
      for (int si = 0; si < allMentions.size(); si++) {
        CorefMention mentions[] = allMentions.get(si);
        Parse p = allParses.get(si);
       
        for (Mention extent : mentionFinder.getMentions(new DefaultParse(p, si))) {
          if (extent.getParse() == null) {
            // not sure how to get head index
            Parse snp = new Parse(p.getText(),extent.getSpan(),"NML",1.0,0);
            p.insert(snp);
          }
        }
       
        Parse tokens[] = p.getTagNodes();
       
        for (CorefMention mention : mentions) {
          Span min = getMinSpan(p, mention);
         
          if (min == null) {
View Full Code Here
112
113
114
115
116
117
118
119
120
121
122
123
  @Override
  protected boolean lastChild(Parse child, Parse parent) {
    boolean lc = super.lastChild(child, parent);
    while(!lc) {
      Parse cp = child.getParent();
      if (cp != parent && cp.getType().equals(child.getType())) {
        lc = super.lastChild(cp,parent);
        child = cp;
      }
      else {
        break;
View Full Code Here
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
      currentChunks[ci].setLabel(Parser.COMPLETE);
      chunks[ci].setLabel(Parser.COMPLETE);
    }
    for (int ci=0;ci<chunks.length;ci++) {
      //System.err.println("parserEventStream.addParseEvents: chunks="+Arrays.asList(chunks));
      Parse parent = chunks[ci].getParent();
      Parse prevParent = chunks[ci];
      int off = 0;
      //build un-built parents
      if (!chunks[ci].isPosTag()) {
        builtNodes.add(off++,chunks[ci]);
      }
      //perform build stages
      while (!parent.getType().equals(AbstractBottomUpParser.TOP_NODE) && parent.getLabel() == null) {
        if (parent.getLabel() == null && !prevParent.getType().equals(parent.getType())) {
          //build level
          if (debug) System.err.println("Build: "+parent.getType()+" for: "+currentChunks[ci]);
          if (etype == ParserEventTypeEnum.BUILD) {
            parseEvents.add(new Event(parent.getType(), buildContextGenerator.getContext(currentChunks, ci)));
          }
          builtNodes.add(off++,parent);
          Parse newParent = new Parse(currentChunks[ci].getText(),currentChunks[ci].getSpan(),parent.getType(),1,0);
          newParent.add(currentChunks[ci],rules);
          newParent.setPrevPunctuation(currentChunks[ci].getPreviousPunctuationSet());
          newParent.setNextPunctuation(currentChunks[ci].getNextPunctuationSet());
          currentChunks[ci].setParent(newParent);
          currentChunks[ci] = newParent;
          newParent.setLabel(Parser.BUILT);
          //see if chunk is complete
          if (lastChild(chunks[ci], parent)) {
            if (etype == ParserEventTypeEnum.CHECK) {
              parseEvents.add(new Event(Parser.COMPLETE, checkContextGenerator.getContext(currentChunks[ci],currentChunks, ci,false)));
            }
            currentChunks[ci].setLabel(Parser.COMPLETE);
            parent.setLabel(Parser.COMPLETE);
          }
          else {
            if (etype == ParserEventTypeEnum.CHECK) {
              parseEvents.add(new Event(Parser.INCOMPLETE, checkContextGenerator.getContext(currentChunks[ci],currentChunks,ci,false)));
            }
            currentChunks[ci].setLabel(Parser.INCOMPLETE);
            parent.setLabel(Parser.COMPLETE);
          }

          chunks[ci] = parent;
          //System.err.println("build: "+newParent+" for "+parent);
        }
        //TODO: Consider whether we need to set this label or train parses at all.
        parent.setLabel(Parser.BUILT);
        prevParent = parent;
        parent = parent.getParent();
      }
      //decide to attach
      if (etype == ParserEventTypeEnum.BUILD) {
        parseEvents.add(new Event(Parser.DONE, buildContextGenerator.getContext(currentChunks, ci)));
      }
      //attach node
      String attachType = null;
      /** Node selected for attachment. */
      Parse attachNode = null;
      int attachNodeIndex = -1;
      if (ci == 0){
        Parse top = new Parse(currentChunks[ci].getText(),new Span(0,currentChunks[ci].getText().length()),AbstractBottomUpParser.TOP_NODE,1,0);
        top.insert(currentChunks[ci]);
      }
      else {
        /** Right frontier consisting of partially-built nodes based on current state of the parse.*/
        List<Parse> currentRightFrontier = Parser.getRightFrontier(currentChunks[0],punctSet);
        if (currentRightFrontier.size() != rightFrontier.size()) {
          System.err.println("fontiers mis-aligned: "+currentRightFrontier.size()+" != "+rightFrontier.size()+" "+currentRightFrontier+" "+rightFrontier);
          System.exit(1);
        }
        Map<Parse, Integer> parents = getNonAdjoinedParent(chunks[ci]);
        //try daughters first.
        for (int cfi=0;cfi<currentRightFrontier.size();cfi++) {
          Parse frontierNode = rightFrontier.get(cfi);
          Parse cfn = currentRightFrontier.get(cfi);
          if (!Parser.checkComplete || !Parser.COMPLETE.equals(cfn.getLabel())) {
            Integer i = parents.get(frontierNode);
            if (debug) System.err.println("Looking at attachment site ("+cfi+"): "+cfn.getType()+" ci="+i+" cs="+nonPunctChildCount(cfn)+", "+cfn+" :for "+currentChunks[ci].getType()+" "+currentChunks[ci]+" -> "+parents);
            if (attachNode == null &&  i != null && i == nonPunctChildCount(cfn)) {
              attachType = Parser.ATTACH_DAUGHTER;
              attachNodeIndex = cfi;
              attachNode = cfn;
              if (etype == ParserEventTypeEnum.ATTACH) {
                parseEvents.add(new Event(attachType, attachContextGenerator.getContext(currentChunks, ci, currentRightFrontier, attachNodeIndex)));
              }
              //System.err.println("daughter attach "+attachNode+" at "+fi);
            }
          }
          else {
            if (debug) System.err.println("Skipping ("+cfi+"): "+cfn.getType()+","+cfn.getPreviousPunctuationSet()+" "+cfn+" :for "+currentChunks[ci].getType()+" "+currentChunks[ci]+" -> "+parents);
          }
          // Can't attach past first incomplete node.
          if (Parser.checkComplete && cfn.getLabel().equals(Parser.INCOMPLETE)) {
            if (debug) System.err.println("breaking on incomplete:"+cfn.getType()+" "+cfn);
            break;
          }
        }
        //try sisters, and generate non-attach events.
        for (int cfi=0;cfi<currentRightFrontier.size();cfi++) {
          Parse frontierNode = rightFrontier.get(cfi);
          Parse cfn = currentRightFrontier.get(cfi);
          if (attachNode == null && parents.containsKey(frontierNode.getParent())
              && frontierNode.getType().equals(frontierNode.getParent().getType())
              ){ //&& frontierNode.getParent().getLabel() == null) {
            attachType = Parser.ATTACH_SISTER;
            attachNode = cfn;
            attachNodeIndex = cfi;
            if (etype == ParserEventTypeEnum.ATTACH) {
              parseEvents.add(new Event(Parser.ATTACH_SISTER, attachContextGenerator.getContext(currentChunks, ci, currentRightFrontier, cfi)));
            }
            chunks[ci].getParent().setLabel(Parser.BUILT);
            //System.err.println("in search sister attach "+attachNode+" at "+cfi);
          }
          else if (cfi == attachNodeIndex) {
            //skip over previously attached daughter.
          }
          else {
            if (etype == ParserEventTypeEnum.ATTACH) {
              parseEvents.add(new Event(Parser.NON_ATTACH, attachContextGenerator.getContext(currentChunks, ci, currentRightFrontier, cfi)));
            }
          }
          //Can't attach past first incomplete node.
          if (Parser.checkComplete && cfn.getLabel().equals(Parser.INCOMPLETE)) {
            if (debug) System.err.println("breaking on incomplete:"+cfn.getType()+" "+cfn);
            break;
          }
        }
        //attach Node
        if (attachNode != null) {
          if (attachType == Parser.ATTACH_DAUGHTER) {
            Parse daughter = currentChunks[ci];
            if (debug) System.err.println("daughter attach a="+attachNode.getType()+":"+attachNode+" d="+daughter+" com="+lastChild(chunks[ci], rightFrontier.get(attachNodeIndex)));
            attachNode.add(daughter,rules);
            daughter.setParent(attachNode);
            if (lastChild(chunks[ci], rightFrontier.get(attachNodeIndex))) {
              if (etype == ParserEventTypeEnum.CHECK) {
                parseEvents.add(new Event(Parser.COMPLETE, checkContextGenerator.getContext(attachNode,currentChunks,ci,true)));
              }
              attachNode.setLabel(Parser.COMPLETE);
            }
            else {
              if (etype == ParserEventTypeEnum.CHECK) {
                parseEvents.add(new Event(Parser.INCOMPLETE, checkContextGenerator.getContext(attachNode,currentChunks,ci,true)));
              }
            }
          }
          else if (attachType == Parser.ATTACH_SISTER) {
            Parse frontierNode = rightFrontier.get(attachNodeIndex);
            rightFrontier.set(attachNodeIndex,frontierNode.getParent());
            Parse sister = currentChunks[ci];
            if (debug) System.err.println("sister attach a="+attachNode.getType()+":"+attachNode+" s="+sister+" ap="+attachNode.getParent()+" com="+lastChild(chunks[ci], rightFrontier.get(attachNodeIndex)));
            Parse newParent = attachNode.getParent().adjoin(sister,rules);

            newParent.setParent(attachNode.getParent());
            attachNode.setParent(newParent);
            sister.setParent(newParent);
            if (attachNode == currentChunks[0]) {
              currentChunks[0]= newParent;
            }
            if (lastChild(chunks[ci], rightFrontier.get(attachNodeIndex))) {
              if (etype == ParserEventTypeEnum.CHECK) {
                parseEvents.add(new Event(Parser.COMPLETE, checkContextGenerator.getContext(newParent,currentChunks,ci,true)));
              }
              newParent.setLabel(Parser.COMPLETE);
            }
            else {
              if (etype == ParserEventTypeEnum.CHECK) {
                parseEvents.add(new Event(Parser.INCOMPLETE, checkContextGenerator.getContext(newParent,currentChunks,ci,true)));
              }
              newParent.setLabel(Parser.INCOMPLETE);
            }

          }
          //update right frontier
          for (int ni=0;ni<attachNodeIndex;ni++) {
View Full Code Here
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
        sentenceNumber=0;
        document.clear();
        parses.clear();
      }
      else {
        Parse p = Parse.parseParse(line);
        parses.add(p);
        Mention[] extents = treebankLinker.getMentionFinder().getMentions(new DefaultParse(p,sentenceNumber));
        //construct new parses for mentions which don't have constituents.
        for (int ei=0,en=extents.length;ei<en;ei++) {
          //System.err.println("PennTreebankLiner.main: "+ei+" "+extents[ei]);

          if (extents[ei].getParse() == null) {
            //not sure how to get head index, but its not used at this point.
            Parse snp = new Parse(p.getText(),extents[ei].getSpan(),"NML",1.0,0);
            p.insert(snp);
            extents[ei].setParse(new DefaultParse(snp,sentenceNumber));
          }

        }
View Full Code Here
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
    }
   
    ParseConverter converter = new ParseConverter(sentence,(Span[])
        tokenSpans.toArray(new Span[tokenSpans.size()]));
  
    Parse unparsedTree = converter.getParseForTagger();
  
    if (unparsedTree.getChildCount() > 0) {
     
      Parse parse = mParser.parse(unparsedTree);
 
      // TODO: We need a strategy to handle the case that a full
      //       parse could not be found. What to do in this case?
     
      parse = converter.transformParseFromTagger(parse);
  
      if (mLogger.isLoggable(Level.INFO)) {
        StringBuffer parseString = new StringBuffer();
        parse.show(parseString);
    
        mLogger.log(Level.INFO, parseString.toString());
      }
  
      createAnnotation(cas, sentenceAnnotation.getBegin(), parse);
View Full Code Here
TOP

Related Classes of opennlp.tools.parser.Parse

  • com.tamingtext.opennlp.ChunkParserTest
  • com.tamingtext.qa.AnswerTypeContextGenerator
  • com.tamingtext.qa.AnswerTypeEventStream
  • com.tamingtext.qa.ChunkParser
  • com.tamingtext.qa.QuestionQParser
  • functionality.Parsing
  • net.sf.nlpshell.CorefParse
  • net.sf.nlpshell.Main
  • opennlp.tools.cmdline.coref.CoreferencerTool
  • opennlp.tools.cmdline.coref.CoreferencerTool$CorefParse
    • 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.