/*
* 修正词性
*/
public static List<Term> updateNature(List<Term> all, Forest... forests) {
if (forests == null) {
if (UserDefineLibrary.FOREST != null) {
forests = new Forest[] {UserDefineLibrary.FOREST};
} else {
return all;
}
}
List<Term> result = new ArrayList<Term>();
for (Term term : all) {
// 添加对正则停用词的支持
if ((stopwordPattern != null) && stopwordPattern.matcher(term.getName()).matches()) {
continue;
}
for (Forest forest : forests) {
String[] params = UserDefineLibrary.getParams(forest, term.getName());
if (params != null) {
term.setNature(new Nature(params[0]));
}
}
result.add(term);
}
return result;
}
/*
* 停用词过滤并且修正词性
*/
public static List<Term> modifResult(List<Term> all) {
List<Term> result = new ArrayList<Term>();
try {
for (Term term : all) {
if (FILTER.size() > 0
&& (FILTER.contains(term.getName())
|| (isTag && FILTER.contains(TAG + term.natrue().natureStr)))) {
continue;
}
// 添加对正则停用词的支持
if ((stopwordPattern != null) && stopwordPattern.matcher(term.getName()).matches()) {
continue;
}
String[] params = UserDefineLibrary.getParams(term.getName());
if (params != null) {
term.setNature(new Nature(params[0]));
}
result.add(term);
}
} catch (Exception e) {
MyStaticValue.LIBRARYLOG.warn(
"FilterStopWord.updateDic can not be null , "
+ "you must use set FilterStopWord.setUpdateDic(map) or use method set map");
}
return result;
}
private void makeNewTerm() {
Term term = new Term(sb.toString(), offe, tempNature.natureStr, 1);
term.selfScore = score;
term.setNature(tempNature);
if (sb.length() > 3) {
term.setSubTerm(TermUtil.getSubTerm(from, to));
}
TermUtil.termLink(from, term);
TermUtil.termLink(term, to);
TermUtil.insertTerm(terms, term);
TermUtil.parseNature(term);
}
public void recognition(Result result) {
List<Term> terms = result.getTerms();
String end = null;
String name;
LinkedList<Term> mergeList = null;
List<Term> list = new LinkedList<Term>();
for (Term term : terms) {
name = term.getName();
if (end == null) {
if ((end = ruleMap.get(name)) != null) {
mergeList = new LinkedList<Term>();
mergeList.add(term);
} else {
list.add(term);
}
} else {
mergeList.add(term);
if (end.equals(name)) {
Term ft = mergeList.pollFirst();
for (Term sub : mergeList) {
ft.merage(sub);
}
ft.setNature(nature);
list.add(ft);
mergeList = null;
end = null;
}
}
}
if (mergeList != null) {
for (Term term : list) {
list.add(term);
}
}
result.setTerms(list);
}
/*
* 停用词过滤并且修正词性
*/
public static List<Term> modifResult(List<Term> all) {
List<Term> result = new ArrayList<Term>();
try {
for (Term term : all) {
if (FILTER.size() > 0
&& (FILTER.contains(term.getName())
|| FILTER.contains(TAG + term.getNatrue().natureStr))) {
continue;
}
String[] params = UserDefineLibrary.getParams(term.getName());
if (params != null) {
term.setNature(new Nature(params[0]));
}
result.add(term);
}
} catch (Exception e) {
System.err.println(
"FilterStopWord.updateDic can not be null , "
+ "you must use set FilterStopWord.setUpdateDic(map) or use method set map");
}
return result;
}