| 孟旭阳,白海燕,吕世炅,宋梦鹏.大模型赋能下学术文献服务中的智能化应用研究[J].情报工程,2025,11(1):003-017 |
| 大模型赋能下学术文献服务中的智能化应用研究 |
| Research on the Intelligent Application of Large Language Models in Academic Literature Retrieval Services |
| |
| DOI： |
| 中文关键词: 大模型；学术文献；智能化应用；科研过程 |
| 英文关键词: Large Language Model; Academic Literature; Intelligent Application; Research Process |
| 基金项目:国家社会科学基金青年项目“大模型赋能下的学术文献智慧服务新模式与应用研究”（24CTQ008）。 |
| 作者 | 单位 | | 孟旭阳 | 中国科学技术信息研究所 北京 100038 | | 白海燕 | 中国科学技术信息研究所 北京 100038 | | 吕世炅 | 中国科学技术信息研究所 北京 100038 | | 宋梦鹏 | 中国科学技术信息研究所 北京 100038 |
|
| 摘要点击次数: 144 |
| 全文下载次数: 31 |
| 中文摘要: |
| [ 目的/ 意义] 大模型等人工智能新技术正影响着人类信息获取和信息创造方式，加速催生学术文献信息服务的新产品、新形态。对学术文献服务中的智能化应用现状进行深入分析，有助于把握现阶段学术文献服务领域智能化应用的成效、不足和未来发展趋势，为进一步的转型升级和智能化创新服务提供参考和启示。[ 方法/ 过程] 在网络调研、文献调研和对各系统实际操作使用的基础上，对国内外典型文献检索发现AI 应用工具的智能应用场景、功能和特点进行总结分析，并与传统学术文献服务进行对比分析，面向科研过程各阶段的科研需求进行支撑分析，对未来学术文献服务的智能化应用发展进行趋势分析和展望。[ 结果/ 结论] 普遍形成了自然语言检索及智能问答的创新服务模式、更细粒度的知识服务内容，在用户输入、信息匹配、结果呈现、交互体验、个性化服务等多个方面初具成效。目前仍处于大模型落地应用初期，结合科研全流程的需求和痛点，提出未来学术文献智能知识服务的展望，一是提升多轮对话下的精准服务能力，向“人— 机— 智”深度融合的高效交互发展，二是打造文本+ 表格+ 图像+ 多维度知识库的细粒度多模态统一智能知识服务，三是个性化、高价值知识创新服务场景与应用的拓展深化。 |
| 英文摘要: |
| [Objective/Significance] Large Language Model (LLM) technologies are reshaping the manner in which humans acquire and create information, expediting the emergence of novel products and formats within academic literature information services. A comprehensive analysis of the current state of intelligent applications in academic literature retrieval services can facilitate an understanding of their effectiveness, limitations, and future development trends in the field of academic literature services at this juncture. Furthermore, such an analysis offers valuable references and inspiration for further transformation,upgrading, and innovative intelligent services. [Methods/Processes] Based on network research, literature review, and practical operation and utilization of various systems, this study summarizes and analyzes the intelligent application scenarios, functions, and characteristics of typical literature retrieval AI application tools, both domestically and internationally. It also conducts a comparative analysis with traditional academic literature retrieval services, supporting the analysis of research needs at all stages of the scientific research process. Furthermore, it undertakes trend analysis and forecasts the future development of intelligent applications in academic literature services. [Results/Conclusions] The innovative service model for natural language retrieval and intelligent question answering, along with more fine-grained knowledge service content, has generally taken shape and has begun to show initial effectiveness in various aspects, including user input, information matching, result presentation, interactive experience, and personalized services. Currently, we are still in the early stages of implementing large-scale models into practical applications. Considering the needs and pain points throughout the entire research process, we propose the following prospects for future intelligent knowledge services in academic literature:Enhancing precision service capabilities in multiturn dialogues and fostering efficient interactions towards the deep integration of “human-machine-intelligence”, establishing a fine-grained, multi-modal unified intelligent knowledge service that integrates text, tables, images, and multi-dimensional knowledge bases, expanding and deepening the scenarios and applications of personalized, high-value knowledge innovation services. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
