用户视角下大语言模型搜索引擎与传统搜索引擎信息检索行为及成效对比研究  

The Comparative Study on the Information Retrieval Behavior and Effectiveness of Large Language Model Search Engine and Traditional Search Engine from the Perspective of Users

作　　者：范雨田 王晓慧[1] Fan Yutian;Wang Xiaohui(School of Management,Liaoning Normal University)

机构地区：[1]辽宁师范大学管理学院

出　　处：《信息与管理研究》2024年第5期24-35,共12页Journal of Information and Management

摘　　要：文章基于用户视角,探究大语言模型搜索引擎与传统搜索引擎在信息检索行为及成效上的差异,以期为大语言模型在文本内容生成与返回机制上的设计与优化提供指导建议,进而提升用户体验和检索效率。结合问卷调查与观察法,本文收集用户在使用大语言模型搜索引擎与传统搜索引擎进行不同检索任务时的行为数据,采用录屏与任务后问卷调查方法,系统记录并分析用户检索行为与成效指标,并运用非参数检验分析,对比不同搜索引擎环境下用户检索行为的差异。研究显示,相较于传统搜索引擎,大语言模型搜索引擎检索式构造长且重构次数少、系统响应虽慢但检索效率高、结果质量评估方面虽显不足但用户满意度较高。研究结果为大语言模型搜索引擎的未来发展与创新路径提供方向性指导,对促进其技术的持续进步具有重要意义。Based on the user’s perspective,this paper explores the differences between the large language model search engine and the traditional search engine in information retrieval behavior and effectiveness,aiming to provide guidance and suggestions for the design and optimization of large language model in text content generation and return mechanism,so as to improve user experience and retrieval efficiency.Combined with questionnaire survey and observation method,this paper collects users’behavior data when using large language model search engine and traditional search engine for different retrieval tasks.Using screen recording and post-task questionnaire survey methods,records and analyzes user retrieval behavior and effectiveness indicators,and uses non-parametric test analysis to compare the differences in user retrieval behavior under different search engine environments.The research shows that compared with the traditional search engine,the large language model search engine has a long retrieval structure and less reconstruction times,the response is slow but the retrieval efficiency is high,and the result quality evaluation is insufficient but the user satisfaction is high.The research results provide directional guidance for the future development and innovation path of large language model search engines,which are of great significance for promoting the continuous technological progress.

关 键 词：大语言模型 传统搜索引擎 信息检索行为 观察法 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术] TP18[自动化与计算机技术—计算机科学与技术]