Examples with Graph com.hankcs.hanlp.seg.common.Graph used on opensource projects

Example 1 with Graph

use of com.hankcs.hanlp.seg.common.Graph in project HanLP by hankcs.

the class NShortSegment method BiOptimumSegment.

List<Vertex> BiOptimumSegment(WordNet wordNetOptimum) {
    //        logger.trace("细分词网：\n{}", wordNetOptimum);
    Graph graph = GenerateBiGraph(wordNetOptimum);
    if (HanLP.Config.DEBUG) {
        System.out.printf("细分词图：%s\n", graph.printByTo());
    }
    NShortPath nShortPath = new NShortPath(graph, 1);
    List<int[]> spResult = nShortPath.getNPaths(1);
    assert spResult.size() > 0 : "最短路径求解失败，请检查下图是否有悬孤节点或负圈\n" + graph.printByTo();
    return graph.parsePath(spResult.get(0));
}

Example 2 with Graph

use of com.hankcs.hanlp.seg.common.Graph in project HanLP by hankcs.

the class NShortSegment method segSentence.

@Override
public List<Term> segSentence(char[] sentence) {
    WordNet wordNetOptimum = new WordNet(sentence);
    WordNet wordNetAll = new WordNet(sentence);
    //        char[] charArray = text.toCharArray();
    // 粗分
    List<List<Vertex>> coarseResult = BiSegment(sentence, 2, wordNetOptimum, wordNetAll);
    boolean NERexists = false;
    for (List<Vertex> vertexList : coarseResult) {
        if (HanLP.Config.DEBUG) {
            System.out.println("粗分结果" + convert(vertexList, false));
        }
        // 实体命名识别
        if (config.ner) {
            wordNetOptimum.addAll(vertexList);
            int preSize = wordNetOptimum.size();
            if (config.nameRecognize) {
                PersonRecognition.Recognition(vertexList, wordNetOptimum, wordNetAll);
            }
            if (config.translatedNameRecognize) {
                TranslatedPersonRecognition.Recognition(vertexList, wordNetOptimum, wordNetAll);
            }
            if (config.japaneseNameRecognize) {
                JapanesePersonRecognition.Recognition(vertexList, wordNetOptimum, wordNetAll);
            }
            if (config.placeRecognize) {
                PlaceRecognition.Recognition(vertexList, wordNetOptimum, wordNetAll);
            }
            if (config.organizationRecognize) {
                // 层叠隐马模型——生成输出作为下一级隐马输入
                vertexList = Dijkstra.compute(GenerateBiGraph(wordNetOptimum));
                wordNetOptimum.addAll(vertexList);
                OrganizationRecognition.Recognition(vertexList, wordNetOptimum, wordNetAll);
            }
            if (!NERexists && preSize != wordNetOptimum.size()) {
                NERexists = true;
            }
        }
    }
    List<Vertex> vertexList = coarseResult.get(0);
    if (NERexists) {
        Graph graph = GenerateBiGraph(wordNetOptimum);
        vertexList = Dijkstra.compute(graph);
        if (HanLP.Config.DEBUG) {
            System.out.printf("细分词网：\n%s\n", wordNetOptimum);
            System.out.printf("细分词图：%s\n", graph.printByTo());
        }
    }
    // 数字识别
    if (config.numberQuantifierRecognize) {
        mergeNumberQuantifier(vertexList, wordNetAll, config);
    }
    // 如果是索引模式则全切分
    if (config.indexMode) {
        return decorateResultForIndexMode(vertexList, wordNetAll);
    }
    // 是否标注词性
    if (config.speechTagging) {
        speechTagging(vertexList);
    }
    if (config.useCustomDictionary) {
        if (config.indexMode)
            combineByCustomDictionary(vertexList, wordNetAll);
        else
            combineByCustomDictionary(vertexList);
    }
    return convert(vertexList, config.offset);
}

Example 3 with Graph

use of com.hankcs.hanlp.seg.common.Graph in project HanLP by hankcs.

the class NShortSegment method BiSegment.

/**
     * 二元语言模型分词
     * @param sSentence 待分词的句子
     * @param nKind     需要几个结果
     * @param wordNetOptimum
     * @param wordNetAll
     * @return 一系列粗分结果
     */
public List<List<Vertex>> BiSegment(char[] sSentence, int nKind, WordNet wordNetOptimum, WordNet wordNetAll) {
    List<List<Vertex>> coarseResult = new LinkedList<List<Vertex>>();
    ////////////////生成词网////////////////////
    GenerateWordNet(wordNetAll);
    //        logger.trace("词网大小：" + wordNetAll.size());
    //        logger.trace("打印词网：\n" + wordNetAll);
    ///////////////生成词图////////////////////
    Graph graph = GenerateBiGraph(wordNetAll);
    //        logger.trace(graph.toString());
    if (HanLP.Config.DEBUG) {
        System.out.printf("打印词图：%s\n", graph.printByTo());
    }
    ///////////////N-最短路径////////////////////
    NShortPath nShortPath = new NShortPath(graph, nKind);
    List<int[]> spResult = nShortPath.getNPaths(nKind * 2);
    if (spResult.size() == 0) {
        throw new RuntimeException(nKind + "-最短路径求解失败，请检查上面的词网是否存在负圈或悬孤节点");
    }
    //////////////日期、数字合并策略
    for (int[] path : spResult) {
        List<Vertex> vertexes = graph.parsePath(path);
        GenerateWord(vertexes, wordNetOptimum);
        coarseResult.add(vertexes);
    }
    return coarseResult;
}