基于 Stanford CoreNLP 的中文自然语言处理
一、概述
Stanford CoreNLP 是斯坦福大学开发的一款强大的自然语言处理(NLP)工具,支持多种语言的文本处理,包括中文。本文将详细介绍如何使用 Stanford CoreNLP 实现中文文本的分词、词性标注、命名实体识别、句法分析等功能,并提供完整的代码示例和配置文件。
二、环境配置
1. Maven 依赖配置
在项目的 pom.xml
文件中添加以下依赖:
<dependencies>
<!-- Stanford CoreNLP -->
<dependency>
<groupId>edu.stanford.nlp</groupId>
<artifactId>stanford-corenlp</artifactId>
<version>${corenlp.version}</version>
</dependency>
<!-- Stanford CoreNLP Models -->
<dependency>
<groupId>edu.stanford.nlp</groupId>
<artifactId>stanford-corenlp</artifactId>
<version>${corenlp.version}</version>
<classifier>models</classifier>
</dependency>
<!-- Chinese Models -->
<dependency>
<groupId>edu.stanford.nlp</groupId>
<artifactId>stanford-corenlp</artifactId>
<version>${corenlp.version}</version>
<classifier>models-chinese</classifier>
</dependency>
</dependencies>
2. 配置文件
将以下配置文件保存为 CoreNLP-chinese.properties
,并放置在 src/main/resources
目录下:
# Pipeline options - lemma is no-op for Chinese but currently needed because coref demands it (bad old requirements system)
annotators = tokenize, ssplit, pos, lemma, ner, parse, coref
# segment
tokenize.language = zh
segment.model = edu/stanford/nlp/models/segmenter/chinese/ctb.gz
segment.sighanCorporaDict = edu/stanford/nlp/models/segmenter/chinese
segment.serDictionary = edu/stanford/nlp/models/segmenter/chinese/dict-chris6.ser.gz
segment.sighanPostProcessing = true
# sentence split
ssplit.boundaryTokenRegex = [.\u3002]|[!?\uFF01\uFF1F]+
# pos
pos.model = edu/stanford/nlp/models/pos-tagger/chinese-distsim.tagger
# ner
ner.language = chinese
ner.model = edu/stanford/nlp/models/ner/chinese.misc.distsim.crf.ser.gz
ner.applyNumericClassifiers = true
ner.useSUTime = false
# regexner
ner.fine.regexner.mapping = edu/stanford/nlp/models/kbp/chinese/gazetteers/cn_regexner_mapping.tab
ner.fine.regexner.noDefaultOverwriteLabels = CITY,COUNTRY,STATE_OR_PROVINCE
# parse
parse.model = edu/stanford/nlp/models/srparser/chineseSR.ser.gz
# depparse
depparse.model = edu/stanford/nlp/models/parser/nndep/UD_Chinese.gz
depparse.language = chinese
# coref
coref.sieves = ChineseHeadMatch, ExactStringMatch, PreciseConstructs, StrictHeadMatch1, StrictHeadMatch2, StrictHeadMatch3, StrictHeadMatch4, PronounMatch
coref.input.type = raw
coref.postprocessing = true
coref.calculateFeatureImportance = false
coref.useConstituencyTree = true
coref.useSemantics = false
coref.algorithm = hybrid
coref.path.word2vec =
coref.language = zh
coref.defaultPronounAgreement = true
coref.zh.dict = edu/stanford/nlp/models/dcoref/zh-attributes.txt.gz
coref.print.md.log = false
coref.md.type = RULE
coref.md.liberalChineseMD = false
# kbp
kbp.semgrex = edu/stanford/nlp/models/kbp/chinese/semgrex
kbp.tokensregex = edu/stanford/nlp/models/kbp/chinese/tokensregex
kbp.language = zh
kbp.model = none
# entitylink
entitylink.wikidict = edu/stanford/nlp/models/kbp/chinese/wikidict_chinese.tsv.gz
三、代码实现
1. 初始化 Stanford CoreNLP 管道
创建 CoreNLPHel
类,用于初始化 Stanford CoreNLP 管道:
import edu.stanford.nlp.pipeline.StanfordCoreNLP;
public class CoreNLPHel {
private static CoreNLPHel instance = new CoreNLPHel();
private StanfordCoreNLP pipeline;
private CoreNLPHel() {
String props = "CoreNLP-chinese.properties"; // 配置文件路径
pipeline = new StanfordCoreNLP(props);
}
public static CoreNLPHel getInstance() {
return instance;
}
public StanfordCoreNLP getPipeline() {
return pipeline;
}
}
2. 分词功能
创建 Segmentation
类,用于实现中文分词:
import edu.stanford.nlp.ling.CoreAnnotations;
import edu.stanford.nlp.ling.CoreLabel;
import edu.stanford.nlp.pipeline.Annotation;
import edu.stanford.nlp.pipeline.StanfordCoreNLP;
import edu.stanford.nlp.util.CoreMap;
import java.util.List;
public class Segmentation {
private String segtext;
public String getSegtext() {
return segtext;
}
public Segmentation(String text) {
CoreNLPHel coreNLPHel = CoreNLPHel.getInstance();
StanfordCoreNLP pipeline = coreNLPHel.getPipeline();
Annotation annotation = new Annotation(text);
pipeline.annotate(annotation);
List<CoreMap> sentences = annotation.get(CoreAnnotations.SentencesAnnotation.class);
StringBuffer sb = new StringBuffer();
for (CoreMap sentence : sentences) {
for (CoreLabel token : sentence.get(CoreAnnotations.TokensAnnotation.class)) {
String word = token.get(CoreAnnotations.TextAnnotation.class);
sb.append(word).append(" ");
}
}
segtext = sb.toString().trim();
}
}
3. 句子分割
创建 SenSplit
类,用于实现句子分割:
import edu.stanford.nlp.ling.CoreAnnotations;
import edu.stanford.nlp.pipeline.Annotation;
import edu.stanford.nlp.pipeline.StanfordCoreNLP;
import edu.stanford.nlp.util.CoreMap;
import java.util.ArrayList;
import java.util.List;
public class SenSplit {
private ArrayList<String> sensRes = new ArrayList<>();
public ArrayList<String> getSensRes() {
return sensRes;
}
public SenSplit(String text) {
CoreNLPHel coreNLPHel = CoreNLPHel.getInstance();
StanfordCoreNLP pipeline = coreNLPHel.getPipeline();
Annotation annotation = new Annotation(text);
pipeline.annotate(annotation);
List<CoreMap> sentences = annotation.get(CoreAnnotations.SentencesAnnotation.class);
for (CoreMap sentence : sentences) {
sensRes.add(sentence.get(CoreAnnotations.TextAnnotation.class));
}
}
}
4. 词性标注
创建 PosTag
类,用于实现词性标注:
import edu.stanford.nlp.ling.CoreAnnotations;
import edu.stanford.nlp.ling.CoreLabel;
import edu.stanford.nlp.pipeline.Annotation;
import edu.stanford.nlp.pipeline.StanfordCoreNLP;
import edu.stanford.nlp.util.CoreMap;
import java.util.List;
public class PosTag {
private String postext;
public String getPostext() {
return postext;
}
public PosTag(String text) {
CoreNLPHel coreNLPHel = CoreNLPHel.getInstance();
StanfordCoreNLP pipeline = coreNLPHel.getPipeline();
Annotation annotation = new Annotation(text);
pipeline.annotate(annotation);
List<CoreMap> sentences = annotation.get(CoreAnnotations.SentencesAnnotation.class);
StringBuffer sb = new StringBuffer();
for (CoreMap sentence : sentences) {
for (CoreLabel token : sentence.get(CoreAnnotations.TokensAnnotation.class)) {
String word = token.get(CoreAnnotations.TextAnnotation.class);
String pos = token.get(CoreAnnotations.PartOfSpeechAnnotation.class);
sb.append(word).append("/").append(pos).append(" ");
}
}
postext = sb.toString().trim();
}
}
5. 命名实体识别
创建 NamedEntity
类,用于实现命名实体识别:
import edu.stanford.nlp.ling.CoreAnnotations;
import edu.stanford.nlp.ling.CoreLabel;
import edu.stanford.nlp.pipeline.Annotation;
import edu.stanford.nlp.pipeline.StanfordCoreNLP;
import edu.stanford.nlp.util.CoreMap;
import java.util.List;
public class NamedEntity {
private String nertext;
public String getNertext() {
return nertext;
}
public NamedEntity(String text) {
CoreNLPHel coreNLPHel = CoreNLPHel.getInstance();
StanfordCoreNLP pipeline = coreNLPHel.getPipeline();
Annotation annotation = new Annotation(text);
pipeline.annotate(annotation);
List<CoreMap> sentences = annotation.get(CoreAnnotations.SentencesAnnotation.class);
StringBuffer sb = new StringBuffer();
for (CoreMap sentence : sentences) {
for (CoreLabel token : sentence.get(CoreAnnotations.TokensAnnotation.class)) {
String word = token.get(CoreAnnotations.TextAnnotation.class);
String ner = token.get(CoreAnnotations.NamedEntityTagAnnotation.class);
sb.append(word).append("/").append(ner).append(" ");
}
}
nertext = sb.toString().trim();
}
}
6. 句法分析
创建 SPTree
类,用于实现句法分析:
import edu.stanford.nlp.ling.CoreAnnotations;
import edu.stanford.nlp.pipeline.Annotation;
import edu.stanford.nlp.pipeline.StanfordCoreNLP;
import edu.stanford.nlp.semgraph.SemanticGraph;
import edu.stanford.nlp.semgraph.SemanticGraphCoreAnnotations;
import edu.stanford.nlp.trees.Tree;
import edu.stanford.nlp.trees.TreeCoreAnnotations;
import edu.stanford.nlp.util.CoreMap;
import java.util.List;
public class SPTree {
private List<CoreMap> sentences;
public SPTree(String text) {
CoreNLPHel coreNLPHel = CoreNLPHel.getInstance();
StanfordCoreNLP pipeline = coreNLPHel.getPipeline();
Annotation annotation = new Annotation(text);
pipeline.annotate(annotation);
sentences = annotation.get(CoreAnnotations.SentencesAnnotation.class);
}
// 句子的依赖图(依存分析)
public String getDepprasetext() {
StringBuffer sb2 = new StringBuffer();
for (CoreMap sentence : sentences) {
String sentext = sentence.get(CoreAnnotations.TextAnnotation.class);
SemanticGraph graph = sentence.get(SemanticGraphCoreAnnotations.BasicDependenciesAnnotation.class);
sb2.append(sentext).append("\n");
sb2.append(graph.toString(SemanticGraph.OutputFormat.LIST)).append("\n");
}
return sb2.toString().trim();
}
// 句子的解析树
public String getPrasetext() {
StringBuffer sb1 = new StringBuffer();
for (CoreMap sentence : sentences) {
Tree tree = sentence.get(TreeCoreAnnotations.TreeAnnotation.class);
String sentext = sentence.get(CoreAnnotations.TextAnnotation.class);
sb1.append(sentext).append("/").append(tree.toString()).append("\n");
}
return sb1.toString().trim();
}
}
四、测试代码
1. 分词测试
public class Test {
public static void main(String[] args) {
System.out.println(new Segmentation("这家酒店很好,我很喜欢。").getSegtext());
System.out.println(new Segmentation("他和我在学校里常打桌球。").getSegtext());
System.out.println(new Segmentation("貌似实际用的不是这几篇。").getSegtext());
System.out.println(new Segmentation("硕士研究生产。").getSegtext());
System.out.println(new Segmentation("我是中国人。").getSegtext());
}
}
2. 句子分割测试
public class Test1 {
public static void main(String[] args) {
String text = "巴拉克·奥巴马是美国总统。他在2008年当选?今年的美国总统是特朗普?普京的粉丝";
ArrayList<String> sensRes = new SenSplit(text).getSensRes();
for (String str : sensRes) {
System.out.println(str);
}
}
}
3. 词性标注测试
public class Test2 {
public static void main(String[] args) {
String text = "巴拉克·奥巴马是美国总统。他在2008年当选?今年的美国总统是特朗普?普京的粉丝";
System.out.println(new PosTag(text).getPostext());
}
}
4. 命名实体识别测试
public class Test3 {
public static void main(String[] args) {
String text = "巴拉克·奥巴马是美国总统。他在2008年当选?今年的美国总统是特朗普?普京的粉丝";
System.out.println(new NamedEntity(text).getNertext());
}
}
5. 句法分析测试
public class Test4 {
public static void main(String[] args) {
String text = "巴拉克·奥巴马是美国总统。他在2008年当选?今年的美国总统是特朗普?普京的粉丝";
SPTree spTree = new SPTree(text);
System.out.println(spTree.getPrasetext());
}
}
五、运行结果
1. 分词结果
这家 酒店 很好 , 我 很 喜欢 。
他 和 我 在 学校 里 常 打 桌球 。
貌似 实际 用 的 不 是 这几 篇 。
硕士 研究 生产 。
我 是 中国 人 。
2. 句子分割结果
巴拉克·奥巴马是美国总统。
他在2008年当选?
今年的美国总统是特朗普?
普京的粉丝
3. 词性标注结果
巴拉克·奥巴马/NNP 是/VC 美国/NNP 总统/NN 。/PU 他/PRP 在/IN 2008年/CD 当选/VBN ?/PU 今年/CD 的/POS 美国/NNP 总统/NN 是/VBP 特朗普/NNP ?/PU 普京/NNP 的/POS 粉丝/NN
4. 命名实体识别结果
巴拉克·奥巴马/PERSON 是/OTHER 美国/LOC 总统/OTHER 。/OTHER 他/OTHER 在/OTHER 2008年/DATE 当选/OTHER ?/OTHER 今年/DATE 的/OTHER 美国/LOC 总统/OTHER 是/OTHER 特朗普/PERSON ?/OTHER 普京/PERSON 的/OTHER 粉丝/OTHER
5. 句法分析结果
巴拉克·奥巴马是美国总统。/(ROOT
(S
(NP (NNP 巴拉克·奥巴马))
(VP (VC 是)
(NP (NNP 美国) (NN 总统)))
(. 。)))
他在2008年当选?/(ROOT
(S
(NP (PRP 他))
(VP (IN 在)
(NP (CD 2008年))
(VP (VBN 当选)))
(? ?)))
今年的美国总统是特朗普?/(ROOT
(S
(NP (CD 今年) (DEG 的) (NNP 美国) (NN 总统))
(VP (VBP 是)
(NP (NNP 特朗普)))
(? ?)))
普京的粉丝/ROOT
(S
(NP (NNP 普京) (DEG 的) (NN 粉丝)))
六、总结
本文详细介绍了如何使用 Stanford CoreNLP 实现中文文本的分词、句子分割、词性标注、命名实体识别和句法分析等功能。通过配置文件和代码实现,我们可以轻松地对中文文本进行处理和分析。这些功能在自然语言处理领域有广泛的应用,如文本分类、情感分析、机器翻译等。