public static double countFoundNouns(AnalysisOutput o) {
int count = 0;
for (int i = 0; i < o.getCNounList().size(); i++) {
if (o.getCNounList().get(i).isExist()) count++;
}
return (count * 100) / o.getCNounList().size();
}
/**
* return the start offset of current decompounds entry.
*
* @param output morphlogical analysis output
* @param index the index of current decompounds entry
* @return the start offset of current decoumpounds entry
*/
private int getStartOffset(AnalysisOutput output, int index) {
int sOffset = 0;
for (int i = 0; i < index; i++) {
sOffset += output.getCNounList().get(i).getWord().length();
}
return sOffset;
}
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);
}
}
}