package com.ruoyi.ai.config;
|
|
import com.ruoyi.ai.store.MongoChatMemoryStore;
|
import dev.langchain4j.memory.chat.ChatMemoryProvider;
|
import dev.langchain4j.memory.chat.MessageWindowChatMemory;
|
import dev.langchain4j.model.embedding.EmbeddingModel;
|
import dev.langchain4j.rag.content.retriever.ContentRetriever;
|
import dev.langchain4j.rag.content.retriever.EmbeddingStoreContentRetriever;
|
import dev.langchain4j.store.embedding.EmbeddingStore;
|
import org.springframework.beans.factory.annotation.Autowired;
|
import org.springframework.context.annotation.Bean;
|
import org.springframework.context.annotation.Configuration;
|
|
/**
|
* AI Agent 配置类
|
* 知识库检索使用数据库管理的向量数据，通过 KnowledgeBaseVector 表管理文件向量记录
|
*
|
* @author :yys
|
* @date : 2025/5/2 20:01
|
*/
|
@Configuration
|
public class XiaozhiAgentConfig {
|
|
@Autowired
|
private MongoChatMemoryStore mongoChatMemoryStore;
|
|
@Autowired
|
private EmbeddingStore embeddingStore;
|
@Autowired
|
private EmbeddingModel embeddingModel;
|
|
@Bean
|
ChatMemoryProvider chatMemoryProviderXiaozhi() {
|
return memoryId -> MessageWindowChatMemory.builder()
|
.id(memoryId)
|
.maxMessages(20)
|
.chatMemoryStore(mongoChatMemoryStore)
|
.build();
|
}
|
|
/**
|
* 知识库内容检索器
|
* 从向量数据库（Pinecone）检索已向量化的知识库内容
|
* 知识库文件通过 KnowledgeBaseVector 表管理，由 KnowledgeRagService 处理向量化
|
*/
|
@Bean
|
ContentRetriever contentRetrieverXiaozhi() {
|
return EmbeddingStoreContentRetriever
|
.builder()
|
.embeddingModel(embeddingModel)
|
.embeddingStore(embeddingStore)
|
.maxResults(1)
|
.minScore(0.8)
|
.build();
|
}
|
}
|
