从 NER 获取全名

Question

通过阅读文档和使用 API，看起来 CoreNLP 会告诉我每个标记的 NER 标签，但它不会帮助我从句子中提取全名。例如：Input: John Wayne and Mary have coffee CoreNLP Output: (John,PERSON) (Wayne,PERSON) (and,O) (Mary,PERSON) (have,O) (coffee,O) Desired Result: list of PERSON ==> [John Wayne, Mary]除非我错过了一些标志，否则我相信要做到这一点，我将需要解析标记并将标记为 PERSON 的连续标记粘合在一起。有人可以确认这确实是我需要做的吗？我主要想知道 CoreNLP 中是否有一些标志或实用程序可以为我做这样的事情。如果有人有实用程序（最好是 Java，因为我使用的是 Java API）可以执行此操作并希望分享，则可加分 :)谢谢！

Answer 1

您可能正在寻找实体提及而不是 NER 标签。例如使用简单 API：

new Sentence("Jimi Hendrix was the greatest").nerTags()

[PERSON, PERSON, O, O, O]

new Sentence("Jimi Hendrix was the greatest").mentions()

[Jimi Hendrix]

StanfordCoreNLP上面的链接有一个使用旧管道的传统非简单 API 的示例

Answer 2

这是完整的 Java API 示例，其中有一个关于实体提及的部分：

import edu.stanford.nlp.coref.data.CorefChain;

import edu.stanford.nlp.ling.*;

import edu.stanford.nlp.ie.util.*;

import edu.stanford.nlp.pipeline.*;

import edu.stanford.nlp.semgraph.*;

import edu.stanford.nlp.trees.*;

import java.util.*;

public class BasicPipelineExample {

  public static String text = "Joe Smith was born in California. " +

      "In 2017, he went to Paris, France in the summer. " +

      "His flight left at 3:00pm on July 10th, 2017. " +

      "After eating some escargot for the first time, Joe said, \"That was delicious!\" " +

      "He sent a postcard to his sister Jane Smith. " +

      "After hearing about Joe's trip, Jane decided she might go to France one day.";

  public static void main(String[] args) {

    // set up pipeline properties

    Properties props = new Properties();

    // set the list of annotators to run

    props.setProperty("annotators", "tokenize,ssplit,pos,lemma,ner,parse,depparse,coref,kbp,quote");

    // set a property for an annotator, in this case the coref annotator is being set to use the neural algorithm

    props.setProperty("coref.algorithm", "neural");

    // build pipeline

    StanfordCoreNLP pipeline = new StanfordCoreNLP(props);

    // create a document object

    CoreDocument document = new CoreDocument(text);

    // annnotate the document

    pipeline.annotate(document);

    // examples

    // 10th token of the document

    CoreLabel token = document.tokens().get(10);

    System.out.println("Example: token");

    System.out.println(token);

    System.out.println();

    // text of the first sentence

    String sentenceText = document.sentences().get(0).text();

    System.out.println("Example: sentence");

    System.out.println(sentenceText);

    System.out.println();

    // second sentence

    CoreSentence sentence = document.sentences().get(1);

    // list of the part-of-speech tags for the second sentence

    List<String> posTags = sentence.posTags();

    System.out.println("Example: pos tags");

    System.out.println(posTags);

    System.out.println();

    // list of the ner tags for the second sentence

    List<String> nerTags = sentence.nerTags();

    System.out.println("Example: ner tags");

    System.out.println(nerTags);

    System.out.println();

    // constituency parse for the second sentence

    Tree constituencyParse = sentence.constituencyParse();

    System.out.println("Example: constituency parse");

    System.out.println(constituencyParse);

    System.out.println();

    // dependency parse for the second sentence

    SemanticGraph dependencyParse = sentence.dependencyParse();

    System.out.println("Example: dependency parse");

    System.out.println(dependencyParse);

    System.out.println();

    // kbp relations found in fifth sentence

    List<RelationTriple> relations =

        document.sentences().get(4).relations();

    System.out.println("Example: relation");

    System.out.println(relations.get(0));

    System.out.println();

    // entity mentions in the second sentence

    List<CoreEntityMention> entityMentions = sentence.entityMentions();

    System.out.println("Example: entity mentions");

    System.out.println(entityMentions);

    System.out.println();

    // coreference between entity mentions

    CoreEntityMention originalEntityMention = document.sentences().get(3).entityMentions().get(1);

    System.out.println("Example: original entity mention");

    System.out.println(originalEntityMention);

    System.out.println("Example: canonical entity mention");

    System.out.println(originalEntityMention.canonicalEntityMention().get());

    System.out.println();

    // get document wide coref info

    Map<Integer, CorefChain> corefChains = document.corefChains();

    System.out.println("Example: coref chains for document");

    System.out.println(corefChains);

    System.out.println();

    // get quotes in document

    List<CoreQuote> quotes = document.quotes();

    CoreQuote quote = quotes.get(0);

    System.out.println("Example: quote");

    System.out.println(quote);

    System.out.println();

    // original speaker of quote

    // note that quote.speaker() returns an Optional

    System.out.println("Example: original speaker of quote");

    System.out.println(quote.speaker().get());

    System.out.println();

    // canonical speaker of quote

    System.out.println("Example: canonical speaker of quote");

    System.out.println(quote.canonicalSpeaker().get());

    System.out.println();

  }

}