-- vector.knowledge_doc_vector definition
-- Drop table
-- DROP TABLE vector.knowledge_doc_vector;
CREATE TABLE vector.knowledge_doc_vector (id bigserial NOT NULL,doc_id bigserial NOT NULL,embedding public.vector NULL,doc_content text NULL,doc_name varchar NULL,doc_page varchar NULL,CONSTRAINT knowledge_doc_vector_pkey PRIMARY KEY (id));

<dependency><groupId>com.baomidou</groupId><artifactId>mybatis-plus-boot-starter</artifactId><version>3.5.3.1</version></dependency> <dependency><groupId>org.postgresql</groupId><artifactId>postgresql</artifactId><version>42.6.0</version></dependency><dependency><groupId>com.pgvector</groupId><artifactId>pgvector</artifactId><version>0.1.6</version></dependency>

import com.baomidou.mybatisplus.core.mapper.BaseMapper;import com.fawvw.pms.vector.domain.dto.KnowledgeDocVectorDTO;import com.fawvw.pms.vector.domain.entity.KnowledgeDocVector;import com.google.gson.Gson;import org.apache.ibatis.annotations.Mapper;import org.apache.ibatis.annotations.SelectProvider;
import java.util.Arrays;import java.util.List;import java.util.Map;
@Mapperpublic interface VectorMapper extends BaseMapper<KnowledgeDocVector> {@SelectProvider(type = SqlProvider.class, method = "findUsersWithSimilarVectors")List<KnowledgeDocVectorDTO> findUsersWithSimilarVectors(double[] targetVector);
static class SqlProvider {public String findUsersWithSimilarVectors(Map<String, Object> param) {System.out.println("param keys:" + new Gson().toJson(param.keySet()));double[] targetVector = (double[]) param.get("array");String sql = "SELECT id, doc_id, embedding, doc_content, " +"cosine_distance(embedding, '" + Arrays.toString(targetVector) + "') AS similarity " +"FROM knowledge_doc_vector " +"ORDER BY similarity DESC";return sql;}}}

public List findSimilarVector(String query) {// query转为向量TextEmbeddingResult embeddingResult = dashCodeHttpApi.textEmbedding(query);List<Double> embeddings = embeddingResult.getOutput().getEmbeddings().get(0).getEmbedding();double[] vector = new double[embeddings.size()];for (int i = 0; i < embeddings.size(); i++) {Double embedding = embeddings.get(i);vector[i] = embedding;}log.info("vector:{}", vector);List<KnowledgeDocVectorDTO> vectorDTOList = vectorMapper.findUsersWithSimilarVectors(vector);log.info("vectorDTOList:{}", vectorDTOList);// 过滤掉相似度低的向量检索结果 阈值:0.7vectorDTOList = vectorDTOList.stream().filter(vectorDTO -> vectorDTO.getSimilarity() <= 0.7).collect(Collectors.toList());
return vectorDTOList;}