图书情报工作

图书情报工作 >

2025 , Vol. 69 >Issue 19: 3 - 12

DOI: https://doi.org/10.13266/j.issn.0252-3116.2025.19.001

研究论文

面向技术主题识别的SAO三元组动态补全方法研究

  • 秦春秀 ,
  • 张从健 ,
  • 吕树月 ,
  • 马续补 ,
  • 王玉龙
展开
  • 1 西安电子科技大学经济与管理学院 西安 710071;
    2 陕西信息资源研究中心 西安 710071
秦春秀，教授，博士，博士生导师，E-mail:cxqin@xidian.edu.cn；张从健，硕士研究生；吕树月，博士研究生；马续补，教授，博士，博士生导师；王玉龙，博士研究生。

收稿日期: 2024-10-21

  修回日期: 2025-02-15

  网络出版日期: 2025-09-12

基金资助

本文系国家社会科学基金重点项目“场景驱动的我国关键核心领域文献资源精细组织与精准服务模式研究”（项目编号：22ATQ002）研究成果之一。

收起

Research on SAO Triplet Dynamic Completion Method for Technology Topic Identification

  • Qin Chunxiu ,
  • Zhang Congjian ,
  • Lü Shuyue ,
  • Ma Xubu ,
  • Wang Yulong
Expand
  • 1 School of Economics and Management, XIDIAN University, Xi'an 710071;
    2 Shaanxi Information Resources Research Center, Xi'an 710071
Qin Chunxiu, Professor, PhD, Doctoral Supervisor, E-mail: cxqin@xidian.edu.cn; Zhang Congjian, Master’s Candidate; Lü Shuyue, Doctoral Candidate; Ma Xubu, Professor, PhD, Doctoral Supervisor; Wang Yulong, Doctoral Candidate.

Received date: 2024-10-21

  Revised date: 2025-02-15

  Online published: 2025-09-12

Supported by

This work is supported by the National Social Science Fund of China project titled “Research on the Fine Organization and Precise Service Mode of Literature Resources in Key Core Areas of China Driven by Scenarios” (Grant No. 22ATQ002).

Fold

摘要

[目的/意义] 识别领域技术主题对于把握领域技术研究态势，为科研管理部门和科研人员提供技术创新决策支持具有重要意义。基于SAO三元组的技术主题识别将主题识别结果从单词层面提升到三元组层面，主题识别的实用性得以提高。然而，该方法仍存在关键技术信息缺失、结果细粒度不高且三元组不易更新等问题。为改善上述问题，提出一种面向技术主题识别的三元组动态补全方法。[方法/过程] 首先，使用OpenIE6对特定领域科技文献摘要数据进行SAO三元组抽取；然后，基于BERTopic对领域SAO三元组进行主题建模，根据三元组的S-A结构识别技术主题三元组，并以该三元组为基本单位进行技术主题表示；最后，通过构建合适的提示词，使用GPT-4o模型结合原始文献摘要动态补全技术主题三元组，从而得出技术语义信息完整的技术主题三元组，进而提升技术主题表示结果的细粒度和可解读性。[结果/结论] 基于高电子迁移率晶体管（HEMT）领域数据进行实证，结果验证提出的方法在领域科技文献技术主题三元组动态补全方面的动态更新能力和提示工程的可靠性，并展现出跨句子级别的补全能力。

关键词： 技术主题识别; SAO三元组; 动态补全; LLM

本文引用格式

秦春秀 , 张从健 , 吕树月 , 马续补 , 王玉龙 . 面向技术主题识别的SAO三元组动态补全方法研究[J]. 图书情报工作, 2025 , 69(19) : 3 -12 . DOI: 10.13266/j.issn.0252-3116.2025.19.001

Abstract

[Purpose/Significance] Identifying technological themes in a given field is crucial for providing decision support for innovation to research management departments and researchers. The technology theme identification based on SAO triplets elevates the results from the word level to the triplet level, enhancing the practicality of theme identification. However, there are still some problems in this method, such as the lack of key technical information, the low granularity of the results and the difficulty in updating triplets. To address these issues, this paper proposes a dynamic completion method for triplets aimed at technology theme identification. [Method/Process] First, SAO triplets were extracted from the abstracts of specific domain literature using OpenIE6. Then, based on BERTopic, the domain SAO triplets were modeled to identify technological theme triplets through their S-A structure, with these triplets serving as the basic unit for representing technological themes. Finally, by constructing appropriate prompts, the GPT-4o model was employed in conjunction with the original abstracts to dynamically complete the technological theme triplets, resulting in triplets with comprehensive semantic information, thus enhancing the granularity and interpretability of the thematic representation. [Result/Conclusion] Based on empirical study conducted on data from the field of high-electron-mobility transistors (HEMTs), the results verify the dynamic updating capability and engineering reliability in terms of dynamic completion of triplets in technical topics within scientific and technological literature, and demonstrate its cross-sentence completion ability.

Key words： technology topic recognition; SAO triplets; dynamic completion; LLM

参考文献

[1] 马铭,王超,周勇,等.基于语义信息的核心技术主题识别与演化趋势分析方法研究[J].情报理论与实践, 2021, 44(9):106-113.(MA M, WANG C, ZHOU Y, et al. Research on core technology topic identification and evolution trend analysis based on semantic information[J]. Information studies:theory & application, 2021, 44(9):106-113.)
[2] 郑德俊,程为.基于三维主题特征测度的新兴主题识别研究[J].情报学报, 2024, 43(2):167-180.(ZHENG D J, CHENG W. Emerging topic recognition based on threedimensional topic feature measurement[J]. Journal of the China Society for Scientific and Technical Information, 2024, 43(2):167-180.)
[3] 刘春江,刘自强,方曙.基于SAO的技术主题创新演化路径识别及其可视化研究[J].情报学报, 2023, 42(2):164-175.(LIU C J, LIU Z Q, FANG S. Evolution path identification and visualization of technological innovation based on SAO[J]. Journal of the China Society for Scientific and Technical Information, 2023, 42(2):164-175.)
[4] HU X M, HONG Z C, ZHANG C W, et al. Reading broadly to open your mind:improving open relation extraction with search documents under self-supervisions[J]. IEEE transactions on knowledge and data engineering, 2024, 36(5):2026-2040.
[5] SHI Y Q, LU Y W, YANG R P, et al. Knowledge graph completion technology research[C]//2022 international conference on computing, robotics and system sciences. Macao:IEEE, 2022:68-73.
[6] CAI B R, XIANG Y, GAO L X, et al. Temporal knowledge graph completion:a survey[C]//Thirty-second international joint conference on artificial intelligence. Macao:IEEE, 2023:6545-6553.
[7] JOAO T A, ELISABETE A, FRANCISCO S, et al. Using dynamic knowledge graphs to detect emerging communities of knowledge[J]. Knowledge-based systems, 2024, 294:111761.
[8] JIANG J H, ZHOU K, DONG Z C, et al. StructGPT:a general framework for large language model to reason over structured data[C]//Conference on empirical methods in natural language processing. Singapore:Association for Computational Linguistics, 2023:9237-9251.
[9] 翟东升,张京先,胡等金.基于SAO结构和词向量的专利技术功效图自动构建研究[J].情报理论与实践, 2020, 43(3):116-123.(ZHAI D S, ZHANG J X, HU D J. Research on automatic construction of patent technology-effect map based on SAO structure and word embedding[J]. Information studies:theory & application, 2020, 43(3):116-123.)
[10] LI R R, WANG X F, LIU Y Q, et al. Improved technology similarity measurement in the medical field based on subjectaction-object semantic structure:a case study of Alzheimer's disease[J]. IEEE transactions on engineering management, 2023, 70(3):280-293.
[11] QIAO Y L, WANG X F, HUANG Y, et al. Tech mining approach for identifying potentially disruptive technologies:from the perspective of technological alternatives[J]. IEEE transactions on engineering management, 2024, 71:5921-5938.
[12] 尹航,李云柯,王志楠,等.科技差距视角下技术机会识别方法优化研究[J].科学学研究, 2025, 43(2):300-310.(YIN H, LI Y K, WANG Z N, et al. Optimization study of technology opportunity identification methods from the perspective of the gap between science and technology[J]. Studies in science of science, 2025, 43(2):300-310.)
[13] 冯立杰,周炜,王金凤,等.基于SAO语义分析的多维技术演化路径与技术创新机会识别研究[J].情报学报, 2022, 41(11):1149-1160.(FENG L J, ZHOU W, WANG J F, et al. Examining multi-dimensional technology evolution path and technology innovation opportunity identification based on SAO semantic analysis[J]. Journal of the China Society for Scientific and Technical Information, 2022, 41(11):1149-1160.)
[14] SHEN T, ZHANG F, CHENG W. A comprehensive overview of knowledge graph completion[J]. Knowledge-based systems, 2022, 255:109597.
[15] WANG Q, MAO Z, WANG B. Knowledge graph embedding:a survey of approaches and applications[J]. IEEE transactions on knowledge and data engineering, 2017, 29(12):2724-2743.
[16] BORDES A, USUNIER N, GARCIA-DURAN A, et al. Translating embeddings for modeling multi-relational data[C]//Proceedings of the 26th international conference on neural information processing systems-Volume 2. Lake Tahoe:Curran Associates, 2013:2787-2795.
[17] 张贞港,余传明.基于实体与关系融合的知识图谱补全模型研究[J].数据分析与知识发现, 2023, 7(2):15-25.(ZHANG Z G, YU C M. Knowledge graph completion model based on entity and relation fusion[J]. Data analysis and knowledge discovery, 2023, 7(2):15-25.)
[18] WANG Z Q, ZHANG Y Y, WANG S D, et al. SINE:secondorder information network embedding[J]. IEEE access, 2020, 8:139044-139051.
[19] 吴越,孙海春.基于图神经网络的知识图谱补全研究综述[J].数据分析与知识发现, 2024, 8(3):10-28.(WU Y, SUN H C. An overview of research on knowledge graph completion based on graph neural network[J]. Data analysis and knowledge discovery, 2024, 8(3):10-28.)
[20] MENG Q, DIT-YAN Y. Temporal link prediction:a unified framework, taxonomy, and review[J]. ACM compute survey, 2024, 56(4):40.
[21] FARGHALY M, MOUNIR M, AREF M, et al. Investigating the challenges and prospects of construction models for dynamic knowledge graphs[J]. IEEE access, 2024, 12:40973-40988.
[22] WEI Y B, HUANG Q S, ZHANG Y, et al. KICGPT:large language model with knowledge in context for knowledge graph completion[C]//Findings of the Association for Computational Linguistics. Singapore:Association for Computational Linguistics, 2023:8667-8683.
[23] OPENAI, ACHIAM J, ADLER S, et al. GPT-4 technical report[DB/OL].[2025-04-12]. https://arxiv.org/abs/2303.08774.
[24] KOLLURU K, ADLAKHA V, AGGARWAL S, et al. OpenIE6:iterative grid labeling and coordination analysis for open information extraction[C]//Proceedings of the 2020 conference on empirical methods in natural language processing. Online:Association for Computational Linguistics, 2020:3748-3761.
[25] GROOTENDORST M. BERTopic:neural topic modeling with a class-based TF-IDF procedure[J]. arXiv preprint, 2022, arXiv:2203.05794.
[26] RATNAYAKE H, WANG C. A prompting framework to enhance language model output[C]//AI 2023:advances in artificial intelligence:36th Australasian joint conference on artificial intelligence, Part II. Berlin:Springer-Verlag, 2023:66-81.
Options
文章导航

/

〈 〉