private float addCount(HashMap<String, Voca> map, String key) {
float value = 1.0f;
if (map.containsKey(key)) {
Voca voca = map.get(key);
value = voca.getCount() + 0.8f;
}
return value;
}
public void wordExtract(HashMap<String, Voca> map, Sentence st) {
System.out.println("# [wordExtract()] 어절의 개수 : " + st.size());
try {
for (int i = 0; i < st.size() - 2; i++) {
Voca voca = new Voca();
String tag = st.get(i).getFirstMorp().getTag();
String fullName = st.get(i).getExp();
String vocaName = st.get(i).getFirstMorp().getString();
String specialTag = st.get(i).get(st.get(i).size() - 1).getTag();
// 체언인지 용언인지 구분
if (isWord(tag)) {
if (fullName.length() < 2) {
continue;
}
if (isSpecial(specialTag)) {
fullName = fullName.substring(0, fullName.length() - 1);
}
voca.setFullName(fullName);
voca.setVocaName(vocaName);
// 다음 단어가 부사와 같은 단어를 걸러내야됨
Eojeol nextEojeol = null;
StringBuffer temp = new StringBuffer();
for (int j = i + 1; j < st.size() - 1; j++) {
nextEojeol = st.get(j);
String nextTag = nextEojeol.getFirstMorp().getTag();
String nextFullName = nextEojeol.getExp();
if (isWord(nextTag)) {
if (nextFullName.length() < 2) {
temp.append(nextFullName + " ");
continue;
}
break;
}
}
String next = "";
if (isSpecial(nextEojeol.get(nextEojeol.size() - 1).getTag())) {
next = nextEojeol.getExp().substring(0, nextEojeol.getExp().length() - 1);
temp.append(next);
voca.setRelatedName(temp.toString());
} else {
temp.append(nextEojeol.getExp());
voca.setRelatedName(temp.toString());
}
String key = fullName + temp.toString();
temp = null;
voca.setCount(addCount(map, key));
map.put(key, voca);
}
}
} catch (Exception e) {
e.printStackTrace();
}
System.out.println("단어 개수 : " + map.size());
}