NLPIR/ICTCLAS for Lucene/Solr 6.6.0 analyzer plugin. Support: MacOS,Linux x86/64, Windows x86/64
The project resources folder is a source folder, which contains all platform's dynamic libraries and push them to the classpath.//Source Folder 保证所有平台下的动态库自动部署到classpath环境下,以便JNA加载动态库。
Lucene-analyzers-nlpir-ictclas is built by Maven. To build Lucene-analyzers-nlpir-ictclas run:
mvn clean package -DskipTests
Or if you use IDE(Eclipse), there is also the same way.
You can use NLPIRTokenizerAnalyzer to do the Chinese Word Segmentation:
- NLPIRTokenizerAnalyzer DEMO
String text="我是中国人";
NLPIRTokenizerAnalyzer nta = new NLPIRTokenizerAnalyzer("", 1, "", "", false);
TokenStream ts = nta.tokenStream("word", text);
ts.reset();
CharTermAttribute term = ts.getAttribute(CharTermAttribute.class);
while(ts.incrementToken()){
System.out.println(term.toString());
}
ts.end();
ts.close();
nta.close();
and also use in Lucene:
- Lucene DEMO
The sample shows how to index your text and search by using NLPIRTokenizerAnalyzer.
//For indexing
NLPIRTokenizerAnalyzer nta = new NLPIRTokenizerAnalyzer("", 1, "", "", false);
IndexWriterConfig inconf=new IndexWriterConfig(nta);
inconf.setOpenMode(OpenMode.CREATE_OR_APPEND);
IndexWriter index=new IndexWriter(FSDirectory.open(Paths.get("index/")),inconf);
Document doc = new Document();
doc.add(new TextField("contents", "特朗普表示,很高兴汉堡会晤后再次同习近平主席通话。我同习主席就重大问题保持沟通和协调、两国加强各层级和各领域交往十分重要。当前,美中关系发展态势良好,我相信可以发展得更好。我期待着对中国进行国事访问。",Field.Store.YES));
index.addDocument(doc);
index.flush();
index.close();
//for searching
String field = "contents";
IndexReader reader = DirectoryReader.open(FSDirectory.open(Paths.get("index/")));
IndexSearcher searcher = new IndexSearcher(reader);
QueryParser parser = new QueryParser(field, nta);
Query query = parser.parse("特朗普习近平");
TopDocs top=searcher.search(query, 100);
ScoreDoc[] hits = top.scoreDocs;
for(int i=0;i<hits.length;i++) {
System.out.println("doc="+hits[i].doc+" score="+hits[i].score);
Document d = searcher.doc(hits[i].doc);
System.out.println(d.get("contents"));
}
To make part of Solr, you need these files:
- the plugin jar, which you have built and put it in your core's lib directory.
- nlpir.properties contains:
data="" #Data directory‘s parent path
encoding=1 #0 GBK;1 UTF-8
sLicenseCode="" # License code
userDict="" # user dictionary, a text file
bOverwrite=false # whether overwrite the existed user dictionary or not
- data directory, you can find it in NLPIR SDK https://github.com/NLPIR-team/NLPIR/tree/master/NLPIR%20SDK/NLPIR-ICTCLAS
Waring: You need to make sure the plugin jar can find the nlpir.properties file. You can put the file to solr_home/server/, and the data need to set the path of NLPIR/ICTCLAS Data.
- Solr Managed-schema
<fieldType name="text_general" class="solr.TextField" positionIncrementGap="100">
<analyzer type="index">
<tokenizer class="org.nlpir.lucene.cn.ictclas.NLPIRTokenizerFactory"/>
</analyzer>
<analyzer type="query">
<tokenizer class="org.nlpir.lucene.cn.ictclas.NLPIRTokenizerFactory"/>
</analyzer>
</fieldType>
- dependency jar for dll: jna.jar. add to your solr's lib.
- v2.*
//Standard Tokenizer
class="org.nlpir.lucene.cn.ictclas.NLPIRTokenizer"
//Finer Segment
class="org.nlpir.lucene.cn.ictclas.finersegmet.FinerTokenizer"
- v1.*
//Standard Tokenizer
class="org.nlpir.lucene.cn.ictclas.NLPIRTokenizer"