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;
}
/**
* 복합명사에서 단위명사를 분리해낸다. 리스트의 가장 마지막에 위치한 단어가 최장단어이다.
*
* @param str 복합명사
* @param pos the analysing start point
* @param o 분석결과 return 단위명사 리스트
* @throws MorphException throw exception
*/
private static List<WordEntry> findNouns(String str, int pos, AnalysisOutput o)
throws MorphException {
List<WordEntry> nList = new ArrayList<WordEntry>();
if (str.length() == 2
&& DictionaryUtil.existSuffix(str.substring(0, 1))
&& DNouns.contains(str.substring(1))) {
o.setStem(o.getStem().substring(0, o.getStem().length() - 1));
o.setNsfx(str.substring(1));
nList.add(new WordEntry(str.substring(0, 1)));
return nList;
} else if (str.length() == 2
&& DictionaryUtil.existSuffix(str.substring(0, 1))
&& DictionaryUtil.existJosa(str.substring(1))) {
return null;
}
if (pos >= 2 && DictionaryUtil.existJosa(str)) return null;
if (str.length() == 1 && (DictionaryUtil.existSuffix(str) || DNouns.contains(str))) {
nList.add(new WordEntry(str));
return nList;
}
for (int i = 1; i < str.length(); i++) {
String sub = str.substring(0, i + 1);
if (!DictionaryUtil.findWithPrefix(sub).hasNext()) break;
WordEntry entry = DictionaryUtil.getAllNoun(sub);
if (entry != null) {
nList.add(entry);
}
}
return nList;
}
/*
* 마지막 음절이 명사형 접미사(등,상..)인지 조사한다.
*/
public static boolean confirmDNoun(AnalysisOutput output) throws MorphException {
int strlen = output.getStem().length();
String d = output.getStem().substring(strlen - 1);
if (!DNouns.contains(d)) return false;
String s = output.getStem().substring(0, strlen - 1);
output.setNsfx(d);
output.setStem(s);
WordEntry cnoun = DictionaryUtil.getAllNoun(s);
if (cnoun != null) {
if (cnoun.getFeature(WordEntry.IDX_NOUN) == '2') output.setCNoun(cnoun.getCompounds());
else output.setCNoun(Collections.EMPTY_LIST);
output.setScore(AnalysisOutput.SCORE_CORRECT);
}
return true;
}
/**
* 용언 + '음/기' + 조사(PTN_VMXMJ)
*
* @param o the analyzed output
* @param candidates candidates
* @throws MorphException throw exception
*/
public static boolean analysisVMJ(AnalysisOutput o, List<AnalysisOutput> candidates)
throws MorphException {
String[] irrs = IrregularUtil.restoreIrregularVerb(o.getStem(), o.getElist().get(0));
if (irrs != null) {
o.setStem(irrs[0]);
o.setElist(irrs[1], 0);
}
if (DictionaryUtil.getVerb(o.getStem()) != null) {
o.setPatn(PatternConstants.PTN_VMJ);
o.setPos(PatternConstants.POS_VERB);
o.setScore(AnalysisOutput.SCORE_CORRECT);
candidates.add(o);
return true;
}
return false;
}
/**
* 어간부가 음/기 로 끝나는 경우
*
* @param o the analyzed output
* @param candidates candidates
* @throws MorphException throw exception
*/
public static boolean analysisMJ(AnalysisOutput o, List<AnalysisOutput> candidates)
throws MorphException {
int strlen = o.getStem().length();
if (strlen < 2) return false;
char[] chrs = MorphUtil.decompose(o.getStem().charAt(strlen - 1));
boolean success = false;
if (o.getStem().charAt(strlen - 1) != '기' && !(chrs.length == 3 && chrs[2] == 'ㅁ'))
return false;
String start = o.getStem();
String end = "";
if (o.getStem().charAt(strlen - 1) == '기') {
start = o.getStem().substring(0, strlen - 1);
end = "기";
} else if (o.getStem().charAt(strlen - 1) == '음') {
start = o.getStem().substring(0, strlen - 1);
end = "음";
}
String[] eomis = EomiUtil.splitEomi(start, end);
if (eomis[0] == null) return false;
String[] pomis = EomiUtil.splitPomi(eomis[0]);
o.setStem(pomis[0]);
o.addElist(eomis[1]);
o.setPomi(pomis[1]);
try {
if (analysisVMJ(o.clone(), candidates)) return true;
if (analysisNSMJ(o.clone(), candidates)) return true;
if (analysisVMXMJ(o.clone(), candidates)) return true;
} catch (CloneNotSupportedException e) {
throw new MorphException(e.getMessage(), e);
}
if (DictionaryUtil.getVerb(o.getStem()) != null) {
o.setPos(PatternConstants.POS_VERB);
o.setPatn(PatternConstants.PTN_VMJ);
o.setScore(AnalysisOutput.SCORE_CORRECT);
candidates.add(o);
return true;
}
return false;
}
public static boolean analysisNSMXMJ(AnalysisOutput o, List<AnalysisOutput> candidates)
throws MorphException {
int idxVbSfix = VerbUtil.endsWithVerbSuffix(o.getStem());
if (idxVbSfix == -1) return false;
o.setVsfx(o.getStem().substring(idxVbSfix));
o.setStem(o.getStem().substring(0, idxVbSfix));
o.setPatn(PatternConstants.PTN_NSMXMJ);
o.setPos(PatternConstants.POS_NOUN);
WordEntry entry = DictionaryUtil.getWordExceptVerb(o.getStem());
if (entry != null) {
if (entry.getFeature(WordEntry.IDX_NOUN) == '0') return false;
else if (o.getVsfx().equals("하") && entry.getFeature(WordEntry.IDX_DOV) != '1') return false;
else if (o.getVsfx().equals("되") && entry.getFeature(WordEntry.IDX_BEV) != '1') return false;
else if (o.getVsfx().equals("내") && entry.getFeature(WordEntry.IDX_NE) != '1') return false;
o.setScore(AnalysisOutput.SCORE_CORRECT); // '입니다'인 경우 인명 등 미등록어가 많이 발생되므로 분석성공으로 가정한다.
} else {
o.setScore(AnalysisOutput.SCORE_ANALYSIS); // '입니다'인 경우 인명 등 미등록어가 많이 발생되므로 분석성공으로 가정한다.
}
candidates.add(o);
return true;
}
/**
* 용언 + '아/어' + 보조용언 + '음/기' + 조사(PTN_VMXMJ)
*
* @param o the analyzed output
* @param candidates candidates
* @throws MorphException throw exception
*/
public static boolean analysisVMXMJ(AnalysisOutput o, List<AnalysisOutput> candidates)
throws MorphException {
int idxXVerb = VerbUtil.endsWithXVerb(o.getStem());
if (idxXVerb != -1) { // 2. 사랑받아보다
String eogan = o.getStem().substring(0, idxXVerb);
o.setXverb(o.getStem().substring(idxXVerb));
String[] stomis = null;
if (eogan.endsWith("아") || eogan.endsWith("어"))
stomis =
EomiUtil.splitEomi(
eogan.substring(0, eogan.length() - 1), eogan.substring(eogan.length() - 1));
else stomis = EomiUtil.splitEomi(eogan, "");
if (stomis[0] == null) return false;
String[] irrs = IrregularUtil.restoreIrregularVerb(stomis[0], stomis[1]);
if (irrs != null) {
o.setStem(irrs[0]);
o.addElist(irrs[1]);
} else {
o.setStem(stomis[0]);
o.addElist(stomis[1]);
}
if (DictionaryUtil.getVerb(o.getStem()) != null) {
o.setPatn(PatternConstants.PTN_VMXMJ);
o.setPos(PatternConstants.POS_VERB);
o.setScore(AnalysisOutput.SCORE_CORRECT);
candidates.add(o);
return true;
} else if (analysisNSMXMJ(o, candidates)) {
return true;
}
}
return false;
}
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);
}
}
}