feat:更新bge-large-zh-v1.5向量模型

Y.Payne

2025-03-23 44b12bca3ac16853bd122cb2dc8f5a8e8163252c

 ruoyi-modules/ruoyi-knowledge/src/main/java/org/ruoyi/knowledge/chain/vectorizer/BgeLargeVectorization.java

¶Ô±ÈÐÂÎļþ
@@ -0,0 +1,63 @@
package org.ruoyi.knowledge.chain.vectorizer;

import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;

import com.google.gson.Gson;
import io.github.ollama4j.OllamaAPI;
import io.github.ollama4j.models.embeddings.OllamaEmbeddingsRequestModel;
import jakarta.annotation.Resource;
import lombok.RequiredArgsConstructor;
import lombok.extern.slf4j.Slf4j;
import org.ruoyi.knowledge.domain.vo.KnowledgeInfoVo;
import org.ruoyi.knowledge.service.IKnowledgeInfoService;
import org.springframework.context.annotation.Lazy;
import org.springframework.stereotype.Component;

@Component
@Slf4j
@RequiredArgsConstructor
public class BgeLargeVectorization implements Vectorization {

    String host = "http://localhost:11434/";

    @Lazy
    @Resource
    private IKnowledgeInfoService knowledgeInfoService;

    @Override
    public List<List<Double>> batchVectorization(List<String> chunkList, String kid) {
        OllamaAPI ollamaAPI = new OllamaAPI(host);
        KnowledgeInfoVo knowledgeInfoVo = knowledgeInfoService.queryById(Long.valueOf(kid));
        List<Double> doubleVector;
        try {
            doubleVector = ollamaAPI.generateEmbeddings(new OllamaEmbeddingsRequestModel(knowledgeInfoVo.getVectorModel(), new Gson().toJson(chunkList)));
        } catch (Exception e) {
            throw new RuntimeException(e);
        }
        List<List<Double>> vectorList = new ArrayList<>();
        vectorList.add(doubleVector);
        return vectorList;
    }

    @Override
    public List<Double> singleVectorization(String chunk, String kid) {
        List<String> chunkList = new ArrayList<>();
        chunkList.add(chunk);
        List<List<Double>> vectorList = batchVectorization(chunkList, kid);
        return vectorList.get(0);
    }

    public static void main(String[] args) {
        OllamaAPI ollamaAPI = new OllamaAPI("http://localhost:11434/");
        List<String> chunkList = Arrays.asList("天很蓝", "海很深");
        List<Double> doubleVector;
        try {
            doubleVector = ollamaAPI.generateEmbeddings(new OllamaEmbeddingsRequestModel("quentinz/bge-large-zh-v1.5", new Gson().toJson(chunkList)));
        } catch (Exception e) {
            throw new RuntimeException(e);
        }
        System.out.println("=== " + doubleVector + " 1===");
    }
}

 ruoyi-modules/ruoyi-knowledge/src/main/java/org/ruoyi/knowledge/chain/vectorizer/VectorizationFactory.java

@@ -1,6 +1,11 @@
package org.ruoyi.knowledge.chain.vectorizer;

import cn.hutool.core.util.StrUtil;
import jakarta.annotation.Resource;
import lombok.extern.slf4j.Slf4j;
import org.ruoyi.knowledge.domain.vo.KnowledgeInfoVo;
import org.ruoyi.knowledge.service.IKnowledgeInfoService;
import org.springframework.context.annotation.Lazy;
import org.springframework.stereotype.Component;

/**
@@ -13,11 +18,28 @@

    private final OpenAiVectorization openAiVectorization;

    public VectorizationFactory(OpenAiVectorization openAiVectorization) {
    private final BgeLargeVectorization bgeLargeVectorization;

    @Lazy
    @Resource
    private IKnowledgeInfoService knowledgeInfoService;

    public VectorizationFactory(OpenAiVectorization openAiVectorization,BgeLargeVectorization bgeLargeVectorization) {
        this.openAiVectorization = openAiVectorization;
        this.bgeLargeVectorization = bgeLargeVectorization;
    }

    public Vectorization getEmbedding(){
        return openAiVectorization;
    public Vectorization getEmbedding(String kid){
        String vectorModel = "text-embedding-3-small";
        if (StrUtil.isNotEmpty(kid)) {
            KnowledgeInfoVo knowledgeInfoVo = knowledgeInfoService.queryById(Long.valueOf(kid));
            if (knowledgeInfoVo != null && StrUtil.isNotEmpty(knowledgeInfoVo.getVectorModel())) {
                vectorModel = knowledgeInfoVo.getVectorModel();
            }
        }
        return switch (vectorModel) {
            case "quentinz/bge-large-zh-v1.5" -> bgeLargeVectorization;
            default -> openAiVectorization;
        };
    }
}

 ruoyi-modules/ruoyi-knowledge/src/main/java/org/ruoyi/knowledge/chain/vectorizer/VectorizationWrapper.java

@@ -1,7 +1,11 @@
package org.ruoyi.knowledge.chain.vectorizer;

import jakarta.annotation.Resource;
import lombok.AllArgsConstructor;
import lombok.extern.slf4j.Slf4j;
import org.ruoyi.knowledge.domain.vo.KnowledgeInfoVo;
import org.ruoyi.knowledge.service.IKnowledgeInfoService;
import org.springframework.context.annotation.Lazy;
import org.springframework.context.annotation.Primary;
import org.springframework.stereotype.Component;

@@ -16,13 +20,13 @@
    private final VectorizationFactory vectorizationFactory;
    @Override
    public List<List<Double>> batchVectorization(List<String> chunkList, String kid) {
        Vectorization embedding = vectorizationFactory.getEmbedding();
        Vectorization embedding = vectorizationFactory.getEmbedding(kid);
        return embedding.batchVectorization(chunkList, kid);
    }

    @Override
    public List<Double> singleVectorization(String chunk, String kid) {
        Vectorization embedding = vectorizationFactory.getEmbedding();
        Vectorization embedding = vectorizationFactory.getEmbedding(kid);
        return embedding.singleVectorization(chunk, kid);
    }
}