摘要
传统招生查询方式问答数据少,构建智能系统难。本文首先用大模型处理招生信息,生成问答集,增强数据后,训练BERT分类问题;其次,用BM25在同类数据中检索;最后,由ChatGLM3生成答案,构建RAG招生问答系统。
The traditional enrollment query method has limited question and answer data,making it difficult to build an intelligent system.This article first uses a large model to process enrollment information,generate a question and answer set,enhance the data,and train BERT classification problems,secondly,BM25 is used to retrieve similar data,finally,ChatGLM3 generates answers and constructs the RAG enrollment question and answer system.
作者
赵耕
涂一雄
杨嘉驰
郑帅
张晴
ZHAO Geng;TU Yixiong;YANG Jiachi;ZHENG Shuai;ZHANG Qing(North China University of Technology,Beijing 100144)
出处
《软件》
2025年第1期175-177,共3页
Software
