private int addBiagramToMap(
String input, int startoffset, Map<String, KoreanToken> map, int position) {
int offset = 0;
int strlen = input.length();
if (strlen < 2) return position;
while (offset < strlen - 1) {
int inc = offset == 0 ? 0 : 1;
if (isAlphaNumChar(input.charAt(offset))) {
String text = findAlphaNumeric(input.substring(offset));
map.put(position + ":" + text, new KoreanToken(text, startoffset + offset, inc));
offset += text.length();
} else {
String text = input.substring(offset, offset + 2 > strlen ? strlen : offset + 2);
map.put(position + ":" + text, new KoreanToken(text, startoffset + offset, inc));
offset++;
}
position += 1;
}
return position - 1;
}
private void extractKeyword(
List<AnalysisOutput> outputs, int startoffset, Map<String, KoreanToken> map, int position) {
int maxDecompounds = 0;
int maxStem = 0;
for (AnalysisOutput output : outputs) {
if (queryMode
&& hasOrigin
&& output.getScore() == AnalysisOutput.SCORE_ANALYSIS
&& output.getCNounList().size() < 2) break;
if (output.getPos() == PatternConstants.POS_VERB) continue; // extract keywords from only noun
if (!originCNoun && output.getCNounList().size() > 0) continue; // except compound nound
int inc = map.size() > 0 ? 0 : 1;
map.put(
position + ":" + output.getStem(), new KoreanToken(output.getStem(), startoffset, inc));
if (output.getStem().length() > maxStem) maxStem = output.getStem().length();
if (output.getCNounList().size() > maxDecompounds)
maxDecompounds = output.getCNounList().size();
// extract the first stem as the keyword for the query processing
if (queryMode) break;
}
if (maxDecompounds > 1) {
for (int i = 0; i < maxDecompounds; i++) {
position += i;
int cPosition = position;
for (AnalysisOutput output : outputs) {
if (output.getPos() == PatternConstants.POS_VERB || output.getCNounList().size() <= i)
continue;
CompoundEntry cEntry = output.getCNounList().get(i);
int cStartoffset = getStartOffset(output, i) + startoffset;
int inc = i == 0 ? 0 : 1;
map.put(
(cPosition) + ":" + cEntry.getWord(),
new KoreanToken(cEntry.getWord(), cStartoffset, inc));
if (bigrammable && !cEntry.isExist())
cPosition = addBiagramToMap(cEntry.getWord(), cStartoffset, map, cPosition);
// extract the words derived from the first stem as the keyword for the query processing
if (queryMode) break;
}
}
} else {
for (AnalysisOutput output : outputs) {
if (output.getPos() == PatternConstants.POS_VERB) continue;
if (bigrammable && output.getScore() < AnalysisOutput.SCORE_COMPOUNDS)
addBiagramToMap(output.getStem(), startoffset, map, position);
}
}
}
/**
* Analyze korean text
*
* @throws MorphException
*/
private void analysisKorean(String input) throws MorphException {
input = trimHangul(input);
List<AnalysisOutput> outputs = morph.analyze(input);
if (outputs.size() == 0) return;
Map<String, KoreanToken> map = new LinkedHashMap<String, KoreanToken>();
if (hasOrigin) map.put("0:" + input, new KoreanToken(input, offsetAtt.startOffset()));
extractKeyword(outputs, offsetAtt.startOffset(), map, 0);
// if(outputs.get(0).getScore()>=AnalysisOutput.SCORE_COMPOUNDS) {
// extractKeyword(outputs,offsetAtt.startOffset(), map, 0);
//// } else {
//// // check whether the input text has some insert spacing errors.
//// List<AnalysisOutput> list = wsAnal.analyze(input);
//// List<AnalysisOutput> results = new ArrayList<AnalysisOutput>();
//// if(list.size()>1 && wsAnal.getOutputScore(list)>AnalysisOutput.SCORE_ANALYSIS) {
//// int offset = 0;
//// for(AnalysisOutput o : list) {
//// if(hasOrigin) map.put(o.getSource(), new
// Token(o.getSource(),offsetAtt.startOffset()+offset,1));
//// results.addAll(morph.analyze(o.getSource()));
//// offset += o.getSource().length();
//// }
//// } else {
//// results.addAll(outputs);
//// }
//// extractKeyword(results, offsetAtt.startOffset(), map, 0);
// }
Collection<KoreanToken> values = map.values();
for (KoreanToken kt : values) {
kt.setOutputs(outputs);
}
morphQueue.addAll(map.values());
}
