基于大模型的外文文献智能检索  

Intelligent Retrieval of Foreign Literature Based on Large Models

作　　者：张保军 ZHANG Baojun(China Electronics Corporation,Beijing 100000)

机构地区：[1]中国电子信息产业集团,北京100000

出　　处：《软件》2025年第2期171-177,共7页Software

摘　　要：传统科技文献检索通常要求用户输入带有逻辑运算符的复杂检索条件,系统转换成ElasticsearchDSL检索式,根据检索式检索并展示结果。外文科技文献的检索情况更为复杂,现代专业领域的外文词汇不断涌现,这些专业词汇往往拼写复杂,且缺乏成熟的中文翻译。尤其是小语种文献,阅读理解和检索难度更高。为此,本文运用人工智能大模型技术,实现智能化、多语种的文献检索,旨在帮助专业人员快速、有效地检索外文文献,省去输入复杂检索式的步骤。研究数据源自国外出版商的开放获取(OA)文献,并可扩展到专利、非OA文献及中文文献领域。整个系统基于大规模微服务的容器化云原生架构和分布式云数据库集群,最终以SaaS形式部署在私有云平台上,为用户提供便捷的访问体验。Traditional scientific literature retrieval typically requires users to input complex search conditions with logical operators,which are then converted into Elasticsearch DSL search equations.Based on the search equations,the system retrieves and displays the results.The retrieval of foreign scientific and technological literature is more complex,and foreign vocabulary in modern professional fields continues to emerge.These professional vocabulary are often spelled complexly and lack mature Chinese translations.This is especially challenging for literature in less commonly spoken languages,which have fewer readers and are even harder to retrieve.This paper leverages large AI models for intelligent multilingual search,aiming to help professionals quickly and effectively retrieve foreign literature in different languages,eliminating the need to enter complex search formulas.The data in this paper is sourced from open access(OA)literature from international publishers,and in practice,the method can also be applied to patents,non-OA literature,and Chinese documents.The entire system is based on advanced distributed cloud database cluster and containerized cloud-native architecture.It is ultimately deployed as a SaaS application on a private cloud platform,making it easy for users to access quickly.

关 键 词：科技文献检索 大模型 人工智能 智能文献检索 云原生 云计算 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]