[{"data":1,"prerenderedAt":58},["Reactive",2],{"site":3,"news-detail-1005":26},{"status":4,"contentType":5,"pagination":6,"data":7},200,"application/json; charset=utf-8",{},{"id":8,"webUrl":9,"mobileUrl":9,"name":10,"locale":11,"dirPath":12,"title":13,"logo":9,"company":14,"address":15,"tel":16,"fax":9,"email":17,"crod":18,"copyright":19,"seoTitle":20,"seoKeyword":21,"seoDescription":22,"sortId":23,"isDefault":8,"status":24,"domains":25},1,null,"default","zh-CN","main","默认站点","成都畅海天科技有限公司","成都市武侯区玉林东路48号","028-67876631","info@codeini.com","蜀ICP备14010005号","版权所有 © 2025-2027 成都畅海天科技有限公司 All Rights Reserved.","Codein 智能体Agent服务平台","Codeini,Agent,互联网开发,智能体，开发，Api,MCP","让每一个 AI 想法，都能变成收入,一小时，从零到可调用的智能",99,0,[],{"model":27,"related":42},{"id":28,"addBy":29,"addTime":30,"categoryTitle":9,"labelTitle":9,"fields":9,"content":31,"categoryRelations":32,"labelRelations":35,"articleAlbums":36,"articleAttachs":37,"siteId":8,"channelId":8,"applicationId":24,"subProgramId":24,"callIndex":38,"articleLocaleKey":9,"title":39,"source":9,"author":9,"linkUrl":9,"imgUrl":9,"videoUrl":9,"seoTitle":9,"seoKeyword":9,"seoDescription":9,"zhaiyao":40,"sortId":34,"click":41,"status":24,"isComment":8,"commentCount":24,"likeCount":24,"contentFormat":24},1005,"admin","2026-06-15 09:00:00","\u003Cp>结合平台 RAG 知识库微服务，分享从\u003Cstrong>文本分块\u003C/strong>、\u003Cstrong>向量检索\u003C/strong>、\u003Cstrong>重排\u003C/strong>到\u003Cstrong>引用溯源\u003C/strong>的工程实践经验。\u003C/p>\u003Cp>合理的分块大小（如 512 tokens）与重叠窗口能显著提升召回质量；检索后引入重排模型，可将最相关片段排在前面，降低无关内容对生成的干扰。\u003C/p>\u003Cp>在回答中标注引用来源片段，不仅提升可信度，也便于用户核验，是企业级知识问答的必备能力。\u003C/p>",[33],{"id":34,"articleId":28,"categoryId":34},5,[],[],[],"news-rag-rerank","RAG 检索重排最佳实践：从召回率到引用溯源","结合平台 RAG 知识库微服务，分享文本分块、向量检索、重排与引用溯源的工程实践经验。",1323,[43],{"id":44,"addBy":29,"addTime":45,"categoryTitle":9,"labelTitle":46,"fields":47,"categoryRelations":48,"labelRelations":51,"articleAlbums":52,"articleAttachs":53,"siteId":8,"channelId":8,"applicationId":24,"subProgramId":24,"callIndex":54,"articleLocaleKey":9,"title":55,"source":9,"author":9,"linkUrl":9,"imgUrl":9,"videoUrl":9,"seoTitle":9,"seoKeyword":9,"seoDescription":9,"zhaiyao":56,"sortId":50,"click":57,"status":24,"isComment":24,"commentCount":24,"likeCount":24,"contentFormat":24},1009,"2026-05-22 09:00:00","",{},[49],{"id":50,"articleId":44,"categoryId":34},9,[],[],[],"news-gateway-arch","架构设计：统一模型网关的适配器模式与负载均衡","深入解析 LLM Bridge 的模型适配器架构，如何屏蔽各家模型 API 差异并支持快速扩展。",1101,1791629298319]