修复learntool部分发现用户自定义次点的问题

kashwy · Mar 4, 2017 · dd34a3b · dd34a3b
1 parent c9dab2c
commit dd34a3b
Show file tree

Hide file tree

Showing 3 changed files with 25 additions and 11 deletions.
diff --git a/pom.xml b/pom.xml
@@ -5,7 +5,7 @@
 	<artifactId>ansj_seg</artifactId>
 	<packaging>jar</packaging>
 	<name>ansj_seg</name>
-	<version>5.1.1</version>
+	<version>5.1.2</version>
 	<description>best java chinese word seg ! </description>
 	<url>https://github.com/NLPchina/ansj_seg</url>
 	<licenses>

diff --git a/src/main/java/org/ansj/dic/LearnTool.java b/src/main/java/org/ansj/dic/LearnTool.java
@@ -7,9 +7,12 @@
 import org.ansj.app.crf.SplitWord;
 import org.ansj.domain.Nature;
 import org.ansj.domain.NewWord;
+import org.ansj.domain.TermNatures;
 import org.ansj.recognition.arrimpl.AsianPersonRecognition;
 import org.ansj.recognition.arrimpl.ForeignPersonRecognition;
+import org.ansj.recognition.impl.NatureRecognition;
 import org.ansj.util.Graph;
+import org.nlpcn.commons.lang.tire.domain.Forest;
 import org.nlpcn.commons.lang.tire.domain.SmartForest;
 import org.nlpcn.commons.lang.util.CollectionUtil;
 
@@ -40,15 +43,22 @@ public class LearnTool {
 	 */
 	private final SmartForest<NewWord> sf = new SmartForest<NewWord>();
 
+	/**
+	 * 学习新词排除用户自定义词典那中的词语
+	 */
+	private Forest[] forests;
+
 	/**
 	 * 公司名称学习.
 	 * 
 	 * @param graph
 	 */
-	public void learn(Graph graph, SplitWord splitWord) {
+	public void learn(Graph graph, SplitWord splitWord, Forest... forests) {
 
 		this.splitWord = splitWord;
 
+		this.forests = forests;
+
 		// 亚洲人名识别
 		if (isAsianName) {
 			findAsianPerson(graph);
@@ -76,7 +86,12 @@ private void addListToTerm(List<NewWord> newWords) {
 		if (newWords.size() == 0)
 			return;
 		for (NewWord newWord : newWords) {
-			addTerm(newWord);
+
+			TermNatures termNatures = new NatureRecognition(forests).getTermNatures(newWord.getName());
+
+			if (termNatures == TermNatures.NULL) {
+				addTerm(newWord);
+			}
 		}
 	}
 
@@ -93,12 +108,12 @@ public void addTerm(NewWord newWord) {
 			temp.update(newWord.getNature(), newWord.getAllFreq());
 		} else {
 			count++;
-			if(splitWord==null){
+			if (splitWord == null) {
 				newWord.setScore(-1);
-			}else{
-				newWord.setScore(-splitWord.cohesion(newWord.getName()));	
+			} else {
+				newWord.setScore(-splitWord.cohesion(newWord.getName()));
 			}
-			
+
 			synchronized (sf) {
 				sf.add(newWord.getName(), newWord);
 			}
@@ -112,8 +127,7 @@ public SmartForest<NewWord> getForest() {
 	/**
 	 * 返回学习到的新词.
 	 * 
-	 * @param num
-	 *            返回数目.0为全部返回
+	 * @param num 返回数目.0为全部返回
 	 * @return
 	 */
 	public List<Entry<String, Double>> getTopTree(int num) {
@@ -138,7 +152,7 @@ public List<Entry<String, Double>> getTopTree(int num, Nature nature) {
 	}
 
 	private void valueResult(SmartForest<NewWord> smartForest, HashMap<String, Double> hm, Nature nature) {
-		
+
 		if (smartForest == null || smartForest.branches == null) {
 			return;
 		}

diff --git a/src/main/java/org/ansj/splitWord/analysis/NlpAnalysis.java b/src/main/java/org/ansj/splitWord/analysis/NlpAnalysis.java
@@ -63,7 +63,7 @@ public List<Term> merger() {
 
 				graph.walkPath();
 
-				learn.learn(graph, splitWord);
+				learn.learn(graph, splitWord ,forests);
 
 				// 姓名识别
 				if (graph.hasPerson && isNameRecognition) {