Examples with CoreMap edu.stanford.nlp.util.CoreMap used on opensource projects

Example 46 with CoreMap

use of edu.stanford.nlp.util.CoreMap in project CoreNLP by stanfordnlp.

the class DocumentPreprocessor method findParagraphSpeaker.

private static String findParagraphSpeaker(Document doc, List<CoreMap> paragraph, int paragraphUtterIndex, String nextParagraphSpeaker, int paragraphOffset, Dictionaries dict) {
    if (!doc.speakers.containsKey(paragraphUtterIndex)) {
        if (!nextParagraphSpeaker.isEmpty()) {
            doc.speakers.put(paragraphUtterIndex, nextParagraphSpeaker);
        } else {
            // cdm [Sept 2015] added this check to try to avoid crash
            if (paragraph.isEmpty()) {
                Redwood.log("debug-preprocessor", "Empty paragraph; skipping findParagraphSpeaker");
                return "";
            }
            CoreMap lastSent = paragraph.get(paragraph.size() - 1);
            String speaker = "";
            boolean hasVerb = false;
            for (int i = 0; i < lastSent.get(CoreAnnotations.TokensAnnotation.class).size(); i++) {
                CoreLabel w = lastSent.get(CoreAnnotations.TokensAnnotation.class).get(i);
                String pos = w.get(CoreAnnotations.PartOfSpeechAnnotation.class);
                String ner = w.get(CoreAnnotations.NamedEntityTagAnnotation.class);
                if (pos.startsWith("V")) {
                    hasVerb = true;
                    break;
                }
                if (ner.startsWith("PER")) {
                    IntTuple headPosition = new IntTuple(2);
                    headPosition.set(0, paragraph.size() - 1 + paragraphOffset);
                    headPosition.set(1, i);
                    if (doc.mentionheadPositions.containsKey(headPosition)) {
                        speaker = Integer.toString(doc.mentionheadPositions.get(headPosition).mentionID);
                    }
                }
            }
            if (!hasVerb && !speaker.equals("")) {
                doc.speakers.put(paragraphUtterIndex, speaker);
            }
        }
    }
    return findNextParagraphSpeaker(doc, paragraph, paragraphOffset, dict);
}

Example 47 with CoreMap

use of edu.stanford.nlp.util.CoreMap in project CoreNLP by stanfordnlp.

the class DocumentPreprocessor method findSpeakersInConversation.

private static void findSpeakersInConversation(Document doc, Dictionaries dict) {
    for (List<Mention> l : doc.predictedMentions) {
        for (Mention m : l) {
            if (m.predicateNominatives == null)
                continue;
            for (Mention a : m.predicateNominatives) {
                if (a.spanToString().toLowerCase().equals("i")) {
                    doc.speakers.put(m.headWord.get(CoreAnnotations.UtteranceAnnotation.class), Integer.toString(m.mentionID));
                }
            }
        }
    }
    List<CoreMap> paragraph = new ArrayList<>();
    int paragraphUtterIndex = 0;
    String nextParagraphSpeaker = "";
    int paragraphOffset = 0;
    for (CoreMap sent : doc.annotation.get(CoreAnnotations.SentencesAnnotation.class)) {
        paragraph.add(sent);
        int currentUtter = sent.get(CoreAnnotations.TokensAnnotation.class).get(0).get(CoreAnnotations.UtteranceAnnotation.class);
        if (paragraphUtterIndex != currentUtter) {
            nextParagraphSpeaker = findParagraphSpeaker(doc, paragraph, paragraphUtterIndex, nextParagraphSpeaker, paragraphOffset, dict);
            paragraphUtterIndex = currentUtter;
            paragraphOffset += paragraph.size();
            paragraph = new ArrayList<>();
        }
    }
    findParagraphSpeaker(doc, paragraph, paragraphUtterIndex, nextParagraphSpeaker, paragraphOffset, dict);
}

Example 48 with CoreMap

use of edu.stanford.nlp.util.CoreMap in project CoreNLP by stanfordnlp.

the class ChineseHcorefDemo method main.

public static void main(String[] args) throws Exception {
    long startTime = System.currentTimeMillis();
    String text = "俄罗斯 航空 公司 一 名 官员 在 ９号 说 ， " + "米洛舍维奇 的 儿子 马可·米洛舍维奇 ９号 早上 持 外交 护照 从 俄国 首都 莫斯科 搭机 飞往 中国 大陆 北京 ， " + "可是 就 在 稍后 就 返回 莫斯科 。 " + "这 名 俄国 航空 公司 官员 说 马可 是 因为 护照 问题 而 在 北京 机场 被 中共 遣返 莫斯科 。 " + "北京 机场 方面 的 这 项 举动 清楚 显示 中共 有意 放弃 在 总统 大选 落败 的 前 南斯拉夫 总统 米洛舍维奇 ， " + "因此 他 在 南斯拉夫 受到 民众 厌恶 的 儿子 马可 才 会 在 北京 机场 被 中共 当局 送回 莫斯科 。 " + "马可 持 外交 护照 能够 顺利 搭机 离开 莫斯科 ， 但是 却 在 北京 受阻 ， 可 算是 踢到 了 铁板 。 " + "可是 这 项 消息 和 先前 外界 谣传 中共 当局 准备 提供 米洛舍维奇 和 他 的 家人 安全 庇护所 有 着 很 大 的 出入 ," + " 一般 认为 在 去年 米洛舍维奇 挥兵 攻打 科索沃 境内 阿尔巴尼亚 一 分离主义 分子 的 时候 ， " + "强力 反对 北约 组织 攻击 南斯拉夫 的 中共 ， 会 全力 保护 米洛舍维奇 和 他 的 家人 及 亲信 。 " + "可是 从 ９号 马可 被 送回 莫斯科 一 事 看 起来 ， 中共 很 可能 会 放弃 米洛舍维奇 。";
    args = new String[] { "-props", "edu/stanford/nlp/hcoref/properties/zh-coref-default.properties" };
    Annotation document = new Annotation(text);
    Properties props = StringUtils.argsToProperties(args);
    StanfordCoreNLP pipeline = new StanfordCoreNLP(props);
    pipeline.annotate(document);
    System.out.println("---");
    System.out.println("coref chains");
    for (CorefChain cc : document.get(CorefCoreAnnotations.CorefChainAnnotation.class).values()) {
        System.out.println("\t" + cc);
    }
    for (CoreMap sentence : document.get(CoreAnnotations.SentencesAnnotation.class)) {
        System.out.println("---");
        System.out.println("mentions");
        for (Mention m : sentence.get(CorefCoreAnnotations.CorefMentionsAnnotation.class)) {
            System.out.println("\t" + m);
        }
    }
    long endTime = System.currentTimeMillis();
    long time = (endTime - startTime) / 1000;
    System.out.println("Running time " + time / 60 + "min " + time % 60 + "s");
}

Example 49 with CoreMap

use of edu.stanford.nlp.util.CoreMap in project CoreNLP by stanfordnlp.

the class CoNLLMentionExtractor method recallErrors.

private static void recallErrors(List<List<Mention>> goldMentions, List<List<Mention>> predictedMentions, Annotation doc) throws IOException {
    List<CoreMap> coreMaps = doc.get(CoreAnnotations.SentencesAnnotation.class);
    int numSentences = goldMentions.size();
    for (int i = 0; i < numSentences; i++) {
        CoreMap coreMap = coreMaps.get(i);
        List<CoreLabel> words = coreMap.get(CoreAnnotations.TokensAnnotation.class);
        Tree tree = coreMap.get(TreeCoreAnnotations.TreeAnnotation.class);
        List<Mention> goldMentionsSent = goldMentions.get(i);
        List<Pair<Integer, Integer>> goldMentionsSpans = extractSpans(goldMentionsSent);
        for (Pair<Integer, Integer> mentionSpan : goldMentionsSpans) {
            logger.finer("RECALL ERROR\n");
            logger.finer(coreMap + "\n");
            for (int x = mentionSpan.first; x < mentionSpan.second; x++) {
                logger.finer(words.get(x).value() + " ");
            }
            logger.finer("\n" + tree + "\n");
        }
    }
}

Example 50 with CoreMap

use of edu.stanford.nlp.util.CoreMap in project CoreNLP by stanfordnlp.

the class CoNLLMentionExtractor method nextDoc.

@Override
public Document nextDoc() throws Exception {
    List<List<CoreLabel>> allWords = new ArrayList<>();
    List<Tree> allTrees = new ArrayList<>();
    CoNLL2011DocumentReader.Document conllDoc = reader.getNextDocument();
    if (conllDoc == null) {
        return null;
    }
    Annotation anno = conllDoc.getAnnotation();
    List<CoreMap> sentences = anno.get(CoreAnnotations.SentencesAnnotation.class);
    for (CoreMap sentence : sentences) {
        if (!Constants.USE_GOLD_PARSES && !replicateCoNLL) {
            // Remove tree from annotation and replace with parse using stanford parser
            sentence.remove(TreeCoreAnnotations.TreeAnnotation.class);
        } else {
            Tree tree = sentence.get(TreeCoreAnnotations.TreeAnnotation.class);
            if (LEMMATIZE) {
                treeLemmatizer.transformTree(tree);
            }
            // generate the dependency graph
            try {
                SemanticGraph deps = SemanticGraphFactory.makeFromTree(tree, SemanticGraphFactory.Mode.ENHANCED, GrammaticalStructure.Extras.NONE);
                SemanticGraph basicDeps = SemanticGraphFactory.makeFromTree(tree, SemanticGraphFactory.Mode.BASIC, GrammaticalStructure.Extras.NONE);
                sentence.set(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation.class, basicDeps);
                sentence.set(SemanticGraphCoreAnnotations.EnhancedDependenciesAnnotation.class, deps);
            } catch (Exception e) {
                logger.log(Level.WARNING, "Exception caught during extraction of Stanford dependencies. Will ignore and continue...", e);
            }
        }
    }
    String preSpeaker = null;
    int utterance = -1;
    for (CoreLabel token : anno.get(CoreAnnotations.TokensAnnotation.class)) {
        if (!token.containsKey(CoreAnnotations.SpeakerAnnotation.class)) {
            token.set(CoreAnnotations.SpeakerAnnotation.class, "");
        }
        String curSpeaker = token.get(CoreAnnotations.SpeakerAnnotation.class);
        if (!curSpeaker.equals(preSpeaker)) {
            utterance++;
            preSpeaker = curSpeaker;
        }
        token.set(CoreAnnotations.UtteranceAnnotation.class, utterance);
    }
    // Run pipeline
    stanfordProcessor.annotate(anno);
    for (CoreMap sentence : anno.get(CoreAnnotations.SentencesAnnotation.class)) {
        allWords.add(sentence.get(CoreAnnotations.TokensAnnotation.class));
        allTrees.add(sentence.get(TreeCoreAnnotations.TreeAnnotation.class));
    }
    // Initialize gold mentions
    List<List<Mention>> allGoldMentions = extractGoldMentions(conllDoc);
    List<List<Mention>> allPredictedMentions;
    if (Constants.USE_GOLD_MENTIONS) {
        //allPredictedMentions = allGoldMentions;
        // Make copy of gold mentions since mentions may be later merged, mentionID's changed and stuff
        allPredictedMentions = makeCopy(allGoldMentions);
    } else if (Constants.USE_GOLD_MENTION_BOUNDARIES) {
        allPredictedMentions = ((RuleBasedCorefMentionFinder) mentionFinder).filterPredictedMentions(allGoldMentions, anno, dictionaries);
    } else {
        allPredictedMentions = mentionFinder.extractPredictedMentions(anno, maxID, dictionaries);
    }
    try {
        recallErrors(allGoldMentions, allPredictedMentions, anno);
    } catch (IOException e) {
        throw new RuntimeException(e);
    }
    Document doc = arrange(anno, allWords, allTrees, allPredictedMentions, allGoldMentions, true);
    doc.conllDoc = conllDoc;
    return doc;
}