1. 簡介

本文介紹如何使用ElasticSearch的Java High Level API執(zhí)行向量查詢，向量類型為dense_vector。

2. ElasticSearch 索引設(shè)計(jì)

PUT caster_vector1
{
  "settings": {
    "number_of_replicas": 0,
    "number_of_shards": 2
  },
  "mappings": {
    "properties": {
      "my_vector": {
        "type": "dense_vector",
        "dims": 2
      },
      "my_text": {
        "type": "text"
      }
    }
  }
}

3. 向量查詢

package com.example.elasticsearchdemo;

import org.apache.http.HttpHost;
import org.elasticsearch.action.search.SearchRequest;
import org.elasticsearch.action.search.SearchResponse;
import org.elasticsearch.client.RequestOptions;
import org.elasticsearch.client.RestClient;
import org.elasticsearch.client.RestHighLevelClient;
import org.elasticsearch.index.query.QueryBuilders;
import org.elasticsearch.index.query.functionscore.ScriptScoreQueryBuilder;
import org.elasticsearch.script.Script;
import org.elasticsearch.script.ScriptType;
import org.elasticsearch.search.SearchHit;
import org.elasticsearch.search.builder.SearchSourceBuilder;

import java.io.IOException;
import java.util.HashMap;
import java.util.Map;

public class ESHighLevelRestQuery {
    public static void main(String[] args) throws IOException {
        // 創(chuàng)建一個(gè)RestHighLevelClient對(duì)象
        RestHighLevelClient client = new RestHighLevelClient(
                RestClient.builder(new HttpHost("192.168.209.3", 9200, "http")));

        // 定義索引名稱和類型名稱
        String indexName = "caster_vector1";
        String typeName = "_doc";
        //定義最低得分
        float minScore = 1.0f;

        //通過分頁限制返回的文檔數(shù)量
        int page = 0;
        int size = 5;

        // 定義查詢向量
        double[] queryVector = new double[]{0, -1};
        // 創(chuàng)建一個(gè)查詢腳本
        Map<String, Object> scriptParams = new HashMap<>();
        scriptParams.put("query_vector", queryVector);

       //指定向量屬性的名字，通過余弦相似度來衡量查詢得分
        Script script = new Script(ScriptType.INLINE, "painless",
                "cosineSimilarity(params.query_vector, 'my_vector') + 1.0", scriptParams);


        // 創(chuàng)建一個(gè)ScriptScoreQueryBuilder對(duì)象
        ScriptScoreQueryBuilder queryBuilder = QueryBuilders.scriptScoreQuery(
                QueryBuilders.matchAllQuery(),
                script);

        // 創(chuàng)建一個(gè)SearchRequest對(duì)象并設(shè)置索引名稱、類型名稱和查詢條件
        SearchRequest searchRequest = new SearchRequest(indexName);
        searchRequest.types(typeName);
        searchRequest.source(new SearchSourceBuilder().query(queryBuilder).from(page).size(size).minScore(minScore));

       // 執(zhí)行搜索操作
        try {
            SearchResponse searchResponse = client.search(searchRequest, RequestOptions.DEFAULT);

            // 處理搜索結(jié)果
            for (SearchHit hit : searchResponse.getHits().getHits()) {
                Map<String, Object> sourceAsMap = hit.getSourceAsMap();
                Float score = hit.getScore();
                System.out.println(sourceAsMap.toString() + ", score: " + score);
            }
        } catch (IOException e) {
            e.printStackTrace();
        }

        // 關(guān)閉RestHighLevelClient對(duì)象
        client.close();

    }
}

在這個(gè)案例中，我們首先創(chuàng)建了一個(gè)RestHighLevelClient對(duì)象，并定義了索引名稱和類型名稱。
為了限制返回的文檔數(shù)量， 我們?cè)O(shè)置了minScore，from，size變量。
然后，定義查詢的向量，創(chuàng)建查詢腳本，腳本中指定向量屬性的名稱，通過余弦相似度來衡量查詢的得分。
接著，我們創(chuàng)建了一個(gè)ScriptScoreQueryBuilder對(duì)象和SearchRequest對(duì)象，執(zhí)行搜索操作，我們將輸出其錯(cuò)誤信息，最后，關(guān)閉了RestHighLevelClient對(duì)象。文章來源地址http://www.zghlxwxcb.cn/news/detail-716857.html

到了這里，關(guān)于ElasticSearch dense_vector向量查詢-Java實(shí)現(xiàn)的文章就介紹完了。如果您還想了解更多內(nèi)容，請(qǐng)?jiān)谟疑辖撬阉鱐OY模板網(wǎng)以前的文章或繼續(xù)瀏覽下面的相關(guān)文章，希望大家以后多多支持TOY模板網(wǎng)！