Revision: 13729
http://gate.svn.sourceforge.net/gate/?rev=13729&view=rev
Author: philgooch
Date: 2011-04-20 18:01:19 +0000 (Wed, 20 Apr 2011)
Log Message:
-----------
Removed lineBreakCount parameter as no longer required (text is now
pre-chunked). Modified processWithMetaMap() to pre-chunk text into chunks
separated by 2 or more line breaks, and strip leading space but keep track of
start offsets for each chunk.
Modified Paths:
--------------
gate/trunk/plugins/Tagger_MetaMap/src/gate/metamap/MetaMapPR.java
Modified: gate/trunk/plugins/Tagger_MetaMap/src/gate/metamap/MetaMapPR.java
===================================================================
--- gate/trunk/plugins/Tagger_MetaMap/src/gate/metamap/MetaMapPR.java
2011-04-20 17:59:22 UTC (rev 13728)
+++ gate/trunk/plugins/Tagger_MetaMap/src/gate/metamap/MetaMapPR.java
2011-04-20 18:01:19 UTC (rev 13729)
@@ -37,6 +37,7 @@
import info.olteanu.interfaces.StringFilter;
import java.util.regex.Pattern;
+import java.util.regex.Matcher;
/**
* This class is the implementation of the resource METAMAP.
@@ -56,10 +57,8 @@
private Boolean annotatePhrases; // annotate MetaMap phrase chunks with
outputASType
private String metaMapOptions; // MetaMap command line string
private Boolean annotateNegEx; // output NegEx results as features to
outputASType
- private Integer linebreakCount; // number of linebreak/whitespace
characters between paragraph chunks
private TaggerMode taggerMode; // term tagger behaviour: all instances,
first only, or first+coreferences
-
-
+
@Override
public Resource init() throws ResourceInstantiationException {
@@ -76,7 +75,6 @@
return this;
}
-
@Override
public void execute() throws ExecutionException {
MetaMapApi mmInst;
@@ -92,9 +90,6 @@
throw new ExecutionException("outputASType parameter must be
set!");
}
- if (linebreakCount == null || !(linebreakCount instanceof Integer)) {
- linebreakCount = 2;
- }
// set up MetaMap API instance
try {
@@ -123,7 +118,7 @@
// once through MetaMap
HashMap<Integer, ArrayList<Integer>> termMapById = new
HashMap<Integer, ArrayList<Integer>>(); // map of annots with content
duplicated by other annots, indexed by annot id
HashMap<String, Integer> termMapByString = new HashMap<String,
Integer>(); // as above, but indexed by string content
-
+
// iterate over each annot of type inputAnn
while (itr.hasNext()) {
@@ -175,8 +170,8 @@
} else {
// string annContent already processed, so keep a
note
// of the annotation id - we'll copy over the
MetaMap annotations later
- Integer uniqueId =
termMapByString.get(annContent);
- ArrayList<Integer> dupAnnIds =
termMapById.get(uniqueId);
+ Integer uniqueId = termMapByString.get(annContent);
+ ArrayList<Integer> dupAnnIds =
termMapById.get(uniqueId);
dupAnnIds.add(annId);
}
} else {
@@ -232,7 +227,7 @@
return;
}
Set keys = termMapById.keySet();
- // System.out.println("Keys: " + keys);
+
Iterator keyIter = keys.iterator();
while (keyIter.hasNext()) {
@@ -245,8 +240,6 @@
fm.put("coreferences:", dupAnnIds);
AnnotationSet mmAnns =
outputAS.getContained(pAnn.getStartNode().getOffset(),
pAnn.getEndNode().getOffset()).get(outputASType);
- //System.out.println(pAnn.getId() + ":" + mmAnns.toString());
-
// Copy all the MetaMap annots within mmAnns to the
// unprocessed annotations that have the same string value
// as the annots we processed through MetaMap
@@ -358,7 +351,6 @@
return api;
}
-
/**
*
* @param pcm - phrase chunk returned by MetaMap
@@ -452,8 +444,6 @@
} // end for
}
-
-
/**
*
* @param pcm - phrase chunk returned by MetaMap
@@ -463,12 +453,12 @@
* @throws Exception
*/
public void processMappings(PCM pcm, List<Negation> negList, Long
lngInitialOffset, Long lngEndOffset) throws Exception {
-
+
List<Mapping> mappings = pcm.getMappingList();
// Mappings are ordered by score, so outputting the first item will
give the mapping
// with the highest score, if requested
- if ( outputMode.equals(OutputMode.HighestMappingOnly) && !
(mappings.isEmpty()) ) {
+ if (outputMode.equals(OutputMode.HighestMappingOnly) &&
!(mappings.isEmpty())) {
processEvents(mappings.get(0).getEvList(), negList, "Mapping",
lngInitialOffset, lngEndOffset);
} else {
for (Mapping map : mappings) {
@@ -477,8 +467,6 @@
}
}
-
-
/**
*
* @param pcm - phrase chunk returned by MetaMap
@@ -528,8 +516,6 @@
} // end for
}
-
-
/**
* @param api - MetaMapApi instance
* @param text - text to be annotated
@@ -543,24 +529,44 @@
*/
List<Result> resultList = null;
- String asciiText = filterNonAscii(normalizeString(text));
+ String asciiText = filterNonAscii(normalizeString(text)) + "\n\n";
- resultList = api.processCitationsFromString(asciiText);
+ // Create a pattern that strips leading whitespace and
+ // and chunks the text delimited by 2 or more blank lines
+ Pattern pattern =
Pattern.compile("(?s)((\\p{Space}|\\p{Cntrl})*)(.+?)\n[\\s]*\n");
- int resultLength = 0;
+ Matcher m = pattern.matcher(asciiText);
+ int iStart = 0;
+ String chunk = "";
+ int chunkLength = 0;
- if (resultList != null) {
- for (Result result : resultList) {
- if (result != null) {
- this.processUtterances(result, lngInitialOffset +
resultLength, (lngEndOffset == null) ? null : lngEndOffset + resultLength);
- // Need to assume linebreakCount line breaks between each
result chunk.
- // The API processCitationsFromString() method chunks text
in linebreak separated fragments,
- // but does not keep track of the offset of each fragment.
- resultLength = resultLength +
result.getInputText().length() + linebreakCount;
- } else {
- throw new Exception("NULL result instance! ");
+ while (m.find(iStart)) {
+ lngInitialOffset += m.group(1).length();
+ chunk = m.group(3);
+ chunkLength = chunk.length();
+
+ if (!chunk.trim().isEmpty()) {
+ resultList = api.processCitationsFromString(chunk);
+
+ int resultLength = 0;
+
+ if (resultList != null) {
+ for (Result result : resultList) {
+ if (result != null) {
+ this.processUtterances(result, lngInitialOffset +
resultLength, (lngEndOffset == null) ? null : lngEndOffset + resultLength);
+ // We've pre-chunked the text now so this loop
should only iterate once
+ // but, if for some reason it iterates more often,
we need to keep track
+ // of the length of the previous result to update
the start offset
+ resultLength = resultLength +
result.getInputText().length();
+ } else {
+ throw new Exception("NULL result instance! ");
+ }
+ }
}
}
+
+ lngInitialOffset += chunkLength;
+ iStart = m.end(3);
}
}
@@ -696,17 +702,7 @@
return annotateNegEx;
}
- @RunTime
- @CreoleParameter(defaultValue = "2",
- comment = "Number of linebreak/whitespace characters between paragraphs")
- public void setLinebreakCount(Integer linebreakCount) {
- this.linebreakCount = linebreakCount;
- }
- public Integer getLinebreakCount() {
- return linebreakCount;
- }
-
@RunTime
@CreoleParameter(defaultValue = "CoReference",
comment = "Map first instance of a term only, first instance plus
coreferences, or all instances independently")
This was sent by the SourceForge.net collaborative development platform, the
world's largest Open Source development site.
------------------------------------------------------------------------------
Benefiting from Server Virtualization: Beyond Initial Workload
Consolidation -- Increasing the use of server virtualization is a top
priority.Virtualization can reduce costs, simplify management, and improve
application availability and disaster protection. Learn more about boosting
the value of server virtualization. http://p.sf.net/sfu/vmware-sfdev2dev
_______________________________________________
GATE-cvs mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/gate-cvs
