elasticsearch 自定义similarity 插件开发

8516830

在搜索开发中，我们要修改打分机制，就需要自定义similarity。现在来简单说一下elasticsearch下的自定义similarity 插件开发。
网上的https://github.com/tlrx/elasticsearch-custom-similarity-provider仅仅支持0.20.0.Beta1-SNAPSHOT版本，现在我们用的版本是elasticsearch 0.90版本以上。那个例子现在不能用，我修改了一下。
1.继承DefaultSimilarity，实现自己的搜索打分机制。

package org.elasticsearch.index.similarity;   import org.apache.lucene.search.similarities.DefaultSimilarity;   /** * Custom similarity class * * @author xq * */ public class CustomSimilarity extends DefaultSimilarity {   @Override public float idf(long docFreq, long numDocs) { return 1.0f; }   }

2.继续AbstractSimilarityProvider，把自定义的打分机制类加载到elasticsearch中。

package org.elasticsearch.index.similarity;   import org.elasticsearch.common.inject.Inject; import org.elasticsearch.common.inject.assistedinject.Assisted; import org.elasticsearch.common.settings.Settings;   /** * Simple {@link SimilarityProvider} for a {@link CustomSimilarity} * * @author xq * */ public class CustomSimilarityProvider extends AbstractSimilarityProvider {   private CustomSimilarity similarity;   @Inject public CustomSimilarityProvider(@Assisted String name, @Assisted Settings settings) { super(name); this.similarity = new CustomSimilarity(); }       public CustomSimilarity get() { return similarity; } }

3.继承AbstractPlugin作为elasticsearch插件使用

public class CustomerSimilarityPlugin extends AbstractPlugin {   @Override public String name() { return "customer-similarity"; }     @Override public String description() { return "customer similarity"; }     @Override public void processModule(Module module) { if (module instanceof SimilarityModule) { SimilarityModule similarityModule = (SimilarityModule) module; similarityModule.addSimilarity("customer-similarity", CustomSimilarityProvider.class); } } }

4.使用

curl -XPOST 'http://host:port/tweeter/' -d ' { "settings": { "similarity": { "index": { "type": "org.elasticsearch.index.similarity.CustomSimilarityProvider" }, "search": { "type": "org.elasticsearch.index.similarity.CustomSimilarityProvider" } } } }'

在创建mapping的使用自定义的打分规则：

{ "news" : { "properties" : { "title" : { "type" : "string", "similarity" : "my_similarity" } } }

在elasticsearch.yml中配置自定义的打分规则类为默认规则。
index.similarity.default.type: my_similarity
相关程序已经放在https://github.com/awnuxkjy/es-custom-similarity-provider,有兴趣的朋友可以参考一下。
把程序打成jar包放在elasticsearch 的plugins 下的 similarity 目录下即可
参考文章：
http://www.elasticsearch.org/guide/reference/index-modules/similarity/
 
http://www.chepoo.com/elasticsearch-similarity-custom-plug-in-development.html