예제 #1
0
파일: APFtoXML.java 프로젝트: rgrishman/jet
 /** generate mention annotations (with entity numbers) based on the ACE entities and mentions. */
 static void addMentionTags(Document doc, AceDocument aceDoc) {
   ArrayList<AceEntity> entities = aceDoc.entities;
   for (int i = 0; i < entities.size(); i++) {
     AceEntity entity = entities.get(i);
     ArrayList<AceEntityMention> mentions = entity.mentions;
     for (int j = 0; j < mentions.size(); j++) {
       AceEntityMention mention = (AceEntityMention) mentions.get(j);
       // we compute a jetSpan not including trailing whitespace
       Span aceSpan = mention.head;
       // skip mentions in ChEnglish APF not aligned to any English text
       if (aceSpan.start() < 0) continue;
       Span jetSpan = new Span(aceSpan.start(), aceSpan.end() + 1);
       FeatureSet features = new FeatureSet("entity", new Integer(i));
       if (flags.contains("types")) {
         features.put("type", entity.type.substring(0, 3));
         if (entity.subtype != null) features.put("subtype", entity.subtype);
       }
       if (flags.contains("extents")) {
         String cleanExtent = mention.text.replaceAll("\n", " ");
         features.put("extent", AceEntityMention.addXmlEscapes(cleanExtent));
       }
       doc.annotate("mention", jetSpan, features);
     }
   }
 }
예제 #2
0
 /**
  * performs the action, adding the specified Annotation. Returns the position of the end of the
  * Annotation.
  */
 @Override
 public int perform(Document doc, PatternApplication patap) {
   Span span;
   HashMap bindings = patap.bestBindings;
   // System.out.println ("bindings (for new annotation): " + bindings);
   if (spanVariable == null) {
     span = new Span(patap.startPosition, patap.bestPosition);
   } else if (spanVariable.name.toString() == "0") {
     span = new Span(patap.startPosition, patap.startPosition);
   } else {
     Object value = bindings.get(spanVariable.name);
     if (value instanceof Span) {
       span = (Span) value;
     } else if (value instanceof Annotation) {
       span = ((Annotation) value).span();
     } else {
       System.out.println("Value of " + spanVariable.toString() + " is not a span.or annotation");
       return -1;
     }
   }
   if (Pat.trace)
     Console.println(
         "Annotating "
             + doc.text(span)
             + " as "
             + type
             + " "
             + features.substitute(bindings).toSGMLString());
   hideAnnotations(doc, type, span);
   hideAnnotations(doc, "token", span);
   Annotation newAnnotation = new Annotation(type, span, features.substitute(bindings));
   doc.addAnnotation(newAnnotation);
   if (bindingVariable != null) bindings.put(bindingVariable.name, newAnnotation);
   return span.end();
 }
예제 #3
0
 /**
  * hides (adds the 'hidden' feature) to all annotations of type <I>type</I> beginning at the
  * starting position of span <I>span</I>.
  */
 public static void hideAnnotations(Document doc, String type, Span span) {
   for (int posn = span.start(); posn < span.end(); posn++) {
     Vector annotations = doc.annotationsAt(posn, type);
     if (annotations != null) {
       for (int i = 0; i < annotations.size(); i++) {
         Annotation ann = (Annotation) annotations.elementAt(i);
         ann.put("hidden", "true");
         // Console.println ("Hiding " + ann);
       }
     }
   }
 }
예제 #4
0
파일: APFtoXML.java 프로젝트: rgrishman/jet
 static void addTimexTags(Document doc, AceDocument aceDoc) {
   List<AceTimex> timeExpressions = aceDoc.timeExpressions;
   for (AceTimex timex : timeExpressions) {
     AceTimexMention mention = (AceTimexMention) timex.mentions.get(0);
     Span aceSpan = mention.extent;
     Span jetSpan = new Span(aceSpan.start(), aceSpan.end() + 1);
     FeatureSet features = new FeatureSet();
     if (timex.val != null && !timex.val.equals("")) features.put("val", timex.val);
     if (timex.anchorVal != null && !timex.anchorVal.equals(""))
       features.put("anchor_val", timex.anchorVal);
     if (timex.anchorDir != null && !timex.anchorDir.equals(""))
       features.put("anchor_dir", timex.anchorDir);
     if (timex.set != null && !timex.set.equals("")) features.put("set", timex.set);
     if (timex.mod != null && !timex.mod.equals("")) features.put("mod", timex.mod);
     doc.annotate("timex2", jetSpan, features);
   }
 }
예제 #5
0
 /** generate mention annotations (with entity numbers) based on the ACE entities and mentions. */
 static void addMentionTags(Document doc, AceDocument aceDoc) {
   ArrayList<AceEntity> entities = aceDoc.entities;
   for (int i = 0; i < entities.size(); i++) {
     AceEntity entity = (AceEntity) entities.get(i);
     ArrayList<AceEntityMention> mentions = entity.mentions;
     for (int j = 0; j < mentions.size(); j++) {
       AceEntityMention mention = mentions.get(j);
       // we compute a jetSpan not including trailing whitespace
       Span aceSpan = mention.head;
       Span jetSpan = new Span(aceSpan.start(), aceSpan.end() + 1);
       FeatureSet features = new FeatureSet("entity", new Integer(i));
       if (showTypes) {
         features.put("type", entity.type.substring(0, 3));
         if (entity.subtype != null) features.put("subtype", entity.subtype);
       }
       doc.annotate("mention", jetSpan, features);
     }
   }
 }
예제 #6
0
파일: APFtoXML.java 프로젝트: rgrishman/jet
 /** write 'fileText' out as file 'XMLfileName' with ENAMEX tags for the names in the document */
 static void addENAMEXtags(Document doc, AceDocument aceDoc) {
   ArrayList<AceEntity> entities = aceDoc.entities;
   for (int i = 0; i < entities.size(); i++) {
     AceEntity entity = entities.get(i);
     ArrayList<AceEntityName> names = entity.names;
     for (int j = 0; j < names.size(); j++) {
       AceEntityName name = names.get(j);
       Span aceSpan = name.extent;
       Span jetSpan = new Span(aceSpan.start(), aceSpan.end() + 1);
       doc.annotate("ENAMEX", jetSpan, new FeatureSet("TYPE", entity.type));
     }
     // for 2004 we have to examine PRE mentions and decide which are names
     if (year.equals("2004")) {
       ArrayList<AceEntityMention> mentions = entity.mentions;
       for (int j = 0; j < mentions.size(); j++) {
         AceEntityMention mention = mentions.get(j);
         String htext = Resolve.normalizeName(mention.headText);
         String[] mentionName = Gazetteer.splitAtWS(htext);
         String preClass = preDict.get(htext.toLowerCase());
         if (mention.type.equals("PRE")) {
           if (gazetteer.isNationality(mentionName)
               || gazetteer.isLocation(mentionName)
               || "N".equals(preClass)) {
             Span aceSpan = mention.head;
             Span jetSpan = new Span(aceSpan.start(), aceSpan.end() + 1);
             doc.annotate("ENAMEX", jetSpan, new FeatureSet("TYPE", entity.type));
           } else if (preClass != null) {
             // do nothing
           } else {
             System.out.println(
                 "Unclassified PRE: " + mention.text + " {" + mention.headText + ")");
             unknownPre.add(htext.toLowerCase());
           }
         }
       }
     }
   }
 }