Examples with TermNatures org.ansj.domain.TermNatures used on opensource projects

Example 1 with TermNatures

use of org.ansj.domain.TermNatures in project ansj_seg by NLPchina.

the class NatureRecognition method getTermNatures.

/**
	 * 传入一次词语获得相关的词性
	 * 
	 * @param word
	 * @return
	 */
public TermNatures getTermNatures(String word) {
    String[] params = null;
    // 获得词性 ， 先从系统辞典。在从用户自定义辞典
    AnsjItem ansjItem = DATDictionary.getItem(word);
    TermNatures tn = null;
    if (ansjItem != AnsjItem.NULL) {
        tn = ansjItem.termNatures;
    } else if ((params = getParams(word)) != null) {
        tn = new TermNatures(new TermNature(params[0], 1));
    } else if (WordAlert.isEnglish(word)) {
        tn = TermNatures.EN;
    } else if (WordAlert.isNumber(word)) {
        tn = TermNatures.M;
    } else {
        tn = TermNatures.NULL;
    }
    return tn;
}

Example 2 with TermNatures

use of org.ansj.domain.TermNatures in project ansj_seg by NLPchina.

the class NatureRecognition method recognition.

/**
	 * 传入一组。词对词语进行。词性标注
	 * 
	 * @param words
	 * @param offe
	 * @return
	 */
public List<Term> recognition(List<String> words, int offe) {
    List<Term> terms = new ArrayList<Term>(words.size());
    int tempOffe = 0;
    for (String word : words) {
        TermNatures tn = getTermNatures(word);
        terms.add(new Term(word, offe + tempOffe, tn));
        tempOffe += word.length();
    }
    new NatureRecognition().recognition(new Result(terms));
    return terms;
}

Example 3 with TermNatures

use of org.ansj.domain.TermNatures in project ansj_seg by NLPchina.

the class Analysis method analysisStr.

/**
	 * 一整句话分词,用户设置的歧异优先
	 * 
	 * @param temp
	 * @return
	 */
private List<Term> analysisStr(String temp) {
    Graph gp = new Graph(temp);
    int startOffe = 0;
    if (this.ambiguityForest != null) {
        GetWord gw = new GetWord(this.ambiguityForest, gp.chars);
        String[] params = null;
        while ((gw.getFrontWords()) != null) {
            if (gw.offe > startOffe) {
                analysis(gp, startOffe, gw.offe);
            }
            params = gw.getParams();
            startOffe = gw.offe;
            for (int i = 0; i < params.length; i += 2) {
                gp.addTerm(new Term(params[i], startOffe, new TermNatures(new TermNature(params[i + 1], 1))));
                startOffe += params[i].length();
            }
        }
    }
    if (startOffe < gp.chars.length) {
        analysis(gp, startOffe, gp.chars.length);
    }
    List<Term> result = this.getResult(gp);
    return result;
}

Example 4 with TermNatures

use of org.ansj.domain.TermNatures in project ansj_seg by NLPchina.

the class Graph method merger.

/**
	 * 具体的遍历打分方法
	 * 
	 * @param i 起始位置
	 * @param j 起始属性
	 * @param to
	 */
private void merger(Term fromTerm, int to, Map<String, Double> relationMap) {
    Term term = null;
    if (terms[to] != null) {
        term = terms[to];
        while (term != null) {
            // 关系式to.set(from)
            term.setPathScore(fromTerm, relationMap);
            term = term.next();
        }
    } else {
        char c = chars[to];
        TermNatures tn = DATDictionary.getItem(c).termNatures;
        if (tn == null || tn == TermNatures.NULL) {
            tn = TermNatures.NULL;
        }
        terms[to] = new Term(String.valueOf(c), to, tn);
        terms[to].setPathScore(fromTerm, relationMap);
    }
}

Example 5 with TermNatures

use of org.ansj.domain.TermNatures in project ansj_seg by NLPchina.

the class NlpAnalysis method getResult.

@Override
protected List<Term> getResult(final Graph graph) {
    Merger merger = new Merger() {

        @Override
        public List<Term> merger() {
            if (learn == null) {
                learn = new LearnTool();
            }
            graph.walkPath();
            learn.learn(graph, splitWord, forests);
            // 姓名识别
            if (graph.hasPerson && isNameRecognition) {
                // 亚洲人名识别
                new AsianPersonRecognition().recognition(graph.terms);
                graph.walkPathByScore();
                NameFix.nameAmbiguity(graph.terms);
                // 外国人名识别
                new ForeignPersonRecognition().recognition(graph.terms);
                graph.walkPathByScore();
            }
            if (splitWord != null) {
                MapCount<String> mc = new MapCount<String>();
                // 通过crf分词
                List<String> words = splitWord.cut(graph.chars);
                Term tempTerm = null;
                int tempOff = 0;
                if (words.size() > 0) {
                    String word = words.get(0);
                    if (!isRuleWord(word)) {
                        mc.add("始##始" + TAB + word, CRF_WEIGHT);
                    }
                }
                for (String word : words) {
                    // 尝试从词典获取词性
                    TermNatures termNatures = new NatureRecognition(forests).getTermNatures(word);
                    Term term = null;
                    if (termNatures != TermNatures.NULL) {
                        term = new Term(word, tempOff, termNatures);
                    } else {
                        term = new Term(word, tempOff, TermNatures.NW);
                        term.setNewWord(true);
                    }
                    // 增加偏移量
                    tempOff += word.length();
                    if (isRuleWord(word)) {
                        // 如果word不对那么不要了
                        tempTerm = null;
                        continue;
                    }
                    if (term.isNewWord()) {
                        // 尝试猜测词性
                        termNatures = NatureRecognition.guessNature(word);
                        term.updateTermNaturesAndNature(termNatures);
                    }
                    TermUtil.insertTerm(graph.terms, term, InsertTermType.SCORE_ADD_SORT);
                    // 对于非词典中的词持有保守态度
                    if (tempTerm != null && !tempTerm.isNewWord() && !term.isNewWord()) {
                        mc.add(tempTerm.getName() + TAB + word, CRF_WEIGHT);
                    }
                    tempTerm = term;
                    if (term.isNewWord()) {
                        learn.addTerm(new NewWord(word, Nature.NW));
                    }
                }
                if (tempTerm != null && !tempTerm.isNewWord()) {
                    mc.add(tempTerm.getName() + TAB + "末##末", CRF_WEIGHT);
                }
                graph.walkPath(mc.get());
            } else {
                LOG.warn("not find any crf model, make sure your config right? ");
            }
            // 数字发现
            if (graph.hasNum && isNumRecognition) {
                new NumRecognition().recognition(graph.terms);
            }
            // 词性标注
            List<Term> result = getResult();
            // 用户自定义词典的识别
            new UserDefineRecognition(InsertTermType.SCORE_ADD_SORT, forests).recognition(graph.terms);
            graph.rmLittlePath();
            graph.walkPathByScore();
            // 进行新词发现
            new NewWordRecognition(learn).recognition(graph.terms);
            graph.walkPathByScore();
            // 优化后重新获得最优路径
            result = getResult();
            // 激活辞典
            for (Term term : result) {
                learn.active(term.getName());
            }
            setRealName(graph, result);
            return result;
        }

        private List<Term> getResult() {
            List<Term> result = new ArrayList<Term>();
            int length = graph.terms.length - 1;
            for (int i = 0; i < length; i++) {
                if (graph.terms[i] == null) {
                    continue;
                }
                result.add(graph.terms[i]);
            }
            return result;
        }
    };
    return merger.merger();
}