图书情报工作

图书情报工作 >

2026 , Vol. 70 >Issue 5: 95 - 103

DOI: https://doi.org/10.13266/j.issn.0252-3116.2026.05.008

研究论文

面向TRIZ的专利技术知识抽取研究

  • 邢思思 ,
  • 钱力
展开
  • 1 中国社会科学院图书馆 北京 100732;
    2 中国科学院文献情报中心 北京 100190;
    3 中国科学院大学经济与管理学院信息资源管理系 北京 100190
邢思思，馆员，硕士；钱力，正高级工程师，博士，博士生导师，通信作者，E-mail：qianl@mail.las.ac.cn。

收稿日期: 2024-12-24

  修回日期: 2025-02-18

  网络出版日期: 2026-03-04

基金资助

本文系国家社会科学基金一般项目“AI4S科技文献知识底座的理论体系及建设方法研究”（项目编号：24BTQ043）研究成果之一。

收起

Research on Patent Technology Knowledge Extraction for TRIZ

  • Xing Sisi ,
  • Qian Li
Expand
  • 1 Chinese Academy of Social Sciences Library, Beijing 100732;
    2 National Science Library, Chinese Academy of Sciences, Beijing 100190;
    3 Department of Information Resources Management, School of Economics and Management, University of Chinese Academy of Sciences, Beijing 100190
Xing Sisi, Librarian, Master; Qian Li, Senior Engineer, PhD, Doctoral Supervisor, Corresponding Author, E-mail: qianl@mail.las.ac.cn.

Received date: 2024-12-24

  Revised date: 2025-02-18

  Online published: 2026-03-04

Supported by

This work is supported by the National Social Science Fund of China project titled “Research on the Theoretical System and Construction Method of the AI4S Knowledge Base of Scientific and Technological Literature” (Grant No.24BTQ043).

Fold

摘要

[目的/意义] 针对专利知识抽取粒度较粗、数据依赖性强、准确率不高等问题，研究面向TRIZ的专利技术知识抽取方法，以提升领域专利知识抽取的深度和准确性。[方法/过程] 在语义TRIZ指导下，面向特定领域延伸细化通用的专利技术要素，基于“预训练—继续预训练—微调”技术路径，构建面向特定领域的BERT-CRF实体识别模型和BERT-softmax关系抽取模型，以薄膜磁头技术领域为例开展实证研究。[结果/结论] 经实证验证，构建的面向薄膜磁头技术领域专利的BERT-MH+CRF实体识别模型F1分数为84.93%，BERT-MH+softmax关系抽取模型F1分数为63.7%。所提出的方法可以有效提升专利知识抽取的效率和效果，为专利语义分析、技术功效矩阵智能构建、领域知识图谱构建等专利技术挖掘场景提供丰富的知识支持。

关键词： 语义TRIZ; 专利文献; 预训练技术; 实体识别; 关系抽取

本文引用格式

邢思思 , 钱力 . 面向TRIZ的专利技术知识抽取研究[J]. 图书情报工作, 2026 , 70(5) : 95 -103 . DOI: 10.13266/j.issn.0252-3116.2026.05.008

Abstract

[Purpose/Significance] To address the issues of coarse granularity, strong data dependency, and low accuracy in patent knowledge extraction, a TRIZ-oriented patent technology knowledge extraction method is studied to enhance its depth and accuracy in specific field. [Method/Process] Under the guidance of semantic TRIZ, general patent technology elements were extended and refined for specific fields. Based on the “pretrain-continual pretrain-fine tuning” method, BERT-CRF entity recognition model and BERT softmax relationship extraction model for specific fields were constructed, and empirical research was conducted using the thin film magnetic head technology field as an example. [Result/Conclusion] After empirical verification, the F1 score of the BERT-MH+CRF entity recognition model for patents in the field of thin-film magnetic head technology is 84.93%, and the F1 score of the BERT-MH+softmax relation extraction model is 63.7%. The method proposed in this paper can effectively improve the efficiency and effect of patent knowledge extraction, and provide rich knowledge support for patent technology mining scenarios such as patent semantic analysis, intelligent construction of technical efficacy matrix, and domain knowledge graph construction.

Key words： semantic TRIZ; patent literature; pre-training techniques; entity recognition; relation extraction

参考文献

[1] 曾蓓, 吕建秋. 基于CiteSpace的国内TRIZ研究热点及前沿分析[J]. 科技管理研究, 2019, 39(18): 260-265. (ZENG B, LV J Q. Hot spot and frontier analysis of domestic TRIZ research based on CiteSpace[J]. Science and technology management research, 2019, 39(18): 260-265.)
[2] 胡正银, 方曙, 张娴, 等. 个性化语义TRIZ构建研究[J]. 图书情报工作, 2015, 59(7): 123-131. (HU Z Y, FANG S, ZHANG X, et al. Study on personalized semantic TRIZ[J]. Library and information service, 2015, 59(7): 123-131.)
[3] KIM Y, TIAN Y S, JEONG Y, et al. Automatic discovery of technology trends from patent text[C]//Proceedings of the 2009 ACM symposium on applied computing. New York: Association for Computing Machinery, 2009: 1480-1487.
[4] CHOI S, KANG D, LIM J, et al. A fact-oriented ontological approach to SAO-based function modeling of patents for implementing function-based technology database[J]. Expert systems with applications, 2012, 39(10): 9129-9140.
[5] GOMILA J M V, MARRO F P. Combining tech-mining and semantic-TRIZ for a faster and better technology analysis: a case in energy storage systems [J]. Technology analysis & strategic management, 2013, 25(6): 725-743.
[6] CHEN D Q. Neural reading comprehension and beyond [D]. Stanford: Stanford University, 2018.
[7] 李卫超. 面向专利的功能信息抽取方法的研究[D]. 天津: 河北工业大学, 2013. (LI W C. The study of function information extraction approach for Chinese patent[D]. Tianjin: Hebei University of Technology, 2013.)
[8] 胡正银, 刘春江, 隗玲, 等. 面向TRIZ的领域专利技术挖掘系统设计与实践[J]. 图书情报工作, 2017, 61(1): 117-124. (HU Z Y, LIU C J, WEI L, et al. Design and practice of domain patent tech mining system oriented to TRIZ[J]. Library and information service, 2017, 61(1): 117-124.)
[9] 赖英旭, 李亚娟, 刘静. 基于本体的水稻育种方法应用知识库构建[J]. 北京工业大学学报, 2019, 45(12): 1181-1191. (LAI Y X, LI Y J, LIU J. Construction of ontology-based rice breeding method knowledge base[J]. Journal of Beijing University of Technology, 2019, 45(12): 1181-1191.)
[10] VERBITSKY M. Semantic TRIZ[J]. TRIZ journal-Espa?ol, 2004: 1-14.
[11] 张娴, 胡正银, 茹丽洁, 等. 专利技术供需信息关联知识组织模式研究[J]. 图书情报工作, 2016, 60(8): 118-125. (ZHANG X, HU Z Y, RU L J, et al. Knowledge organization framework for the association of patent technology supply and demand information[J]. Library and information service, 2016, 60(8): 118-125.)
[12] CHOI S, KIM H, YOON J, et al. An SAO-based text-mining approach for technology road mapping using patent information[J]. Management of innovation eJournal, 2013, 43(1): 52-74.
[13] WANG X F, QIU P J, ZHU D H, et al. Identification of technology development trends based on subject-action-object analysis: the case of dye-sensitized solar cells[J]. Technological forecasting and social change, 2015(98): 24-46.
[14] 李晓曼, 张学福, 宋红燕, 等. 专利文献技术要素识别方法研究——以纳米肥料领域为例[J]. 图书情报工作, 2020, 64(6): 59-68. (LI X M, ZHANG X F, SONG H Y, et al. Research on selection and identification of technical elements for patent technological evolution analysis: a case study on nano fertilizer[J]. Library and information service, 2020, 64(6): 59-68.)
[15] 刘春江, 李姝影, 方曙, 等. 面向TRIZ的专利技术三元组抽取研究与应用[J]. 数据分析与知识发现, 2024, 8(6): 158-167. (LIU C J, LI S Y, FANG S, et al. Extracting triplets of technology patents for TRIZ[J]. Data analysis and knowledge discovery, 2024, 8(6): 158-167.)
[16] 马建红, 张明月, 赵亚男. 面向创新设计的专利知识抽取方法[J]. 计算机应用, 2016, 36(2): 465-471. (MA J H, ZHANG M Y, ZHAO Y N. Patent knowledge extraction method for innovation design[J]. Journal of computer applications, 2016, 36(2): 465-471.)
[17] NANBA H, KONDO T, TAKEZAWA T. Automatic creation of a technical trend map from research papers and patents[C]//Proceedings of the 3rd international workshop on Patent information retrieval. New York: Association for Computing Machinery, 2010: 11-16.
[18] 张颖怡, 章成志. 基于学术论文全文的研究方法句自动抽取研究[J]. 情报学报, 2020, 39(6): 640-650. (ZHANG Y Y, XHANG C Z. Methodological and automatic sentence extraction from academic article’s full-text[J]. Journal of the China Society for Scientific and Technical Information, 2020, 39(6): 640-650.)
[19] PETERS M E, NEUMANN M, IYYER M, et al. Deep contextualized word representations[C]//Proceedings of the 2018 conference of the north American chapter of the association for computational linguistics: human language technologies. Stroudsburg: Association for Computational Linguistics, 2018: 2227-2237.
[20] RADFORD A, NARASIMHAN K, SALIMANS T, et al. Improving language understanding by generative pre-training[J]. 2018.
[21] DEVLIN J, CHANG M W, LEE K, et al. Bert: pre-training of deep bidirectional transformers for language understanding[C]//Conference of the north American chapter of the association for computational linguistics: human language technologies, Minneapolis: Association for Computational Linguistics, 2018: 4171-4186.
[22] SAAD F, ARAS H, HACKL-SOMMER R. Improving named entity recognition for biomedical and patent data using Bi-LSTM deep neural network models[M]. Natural language processing and information systems. Saarbrücken: Springer International Publishing, 2020: 25-36.
[23] ZHANG C Z, MAYR P, LU W, et al. JCDL2022 workshop: extraction and evaluation of knowledge entities from scientific documents (EEKE2022) [C]//Proceedings of the 22nd ACM/IEEE joint conference on digital libraries, New York: Association for Computing Machinery, 2022: 1-2.
[24] 曹树金, 李睿婧.基于专利文献摘要的创新知识图谱构建与应用[J]. 情报理论与实践, 2022, 45(11): 21-28. (CAO S J, LI R J. The construction and application of innovation knowledge graph for patent documents abstract[J]. Information studies: theory & application, 2022, 45(11): 21-28.)
[25] 何喜军, 张佑, 孟雪, 等. 专利供需知识图谱半自动化构建及应用[J]. 情报杂志, 2023, 42(3): 139-150. (HE X J, ZHANG Y, MENG X, et al. Semi-automatic construction and application scenarios of PSD-KG[J]. Journal of intelligence, 2023, 42(3): 139-150.)
[26] ZHOU H Y, CHEN X Y, ZHANG Y H, et al. Generalized radiograph representation learning via cross-supervision between images and freetext radiology reports[J]. Nature Machine Intelligence, 2022(4): 32-40.
[27] LEE J, YOON W, KIM S, et al. BioBERT: a pre-trained biomedical language representation model for biomedical text mining[J]. Bioinformatics, 2020, 36(4): 1234-1240.
[28] GUPTA T, ZAKI M, ANOOP KRISHNAN N M, et al. 2022. MatSciBERT: a materials domain language model for text mining and information extraction[J]. NPJ computational materials, 2022(8): 1-11.
[29] 谭晶维, 张怀清, 刘洋, 等. 问答式林业预训练语言模型ForestBERT[J]. 林业科学, 2024, 60(9): 99-110. (TAN Q W, ZHANG H Q, LIU Y, et al. Question-answering forestry pre-trained language model: ForestBERT[J]. Scientia silvae sinicae, 2024, 60(9): 99-110.)
[30] HU S, ZHANG H, HU X, et al. Chinese named entity recognition based on BERT-CRF model[C]//2022 IEEE/ACIS 22nd international conference on computer and information science (ICIS). Zhuhai: IEEE, 2022: 105-108.
[31] SOARES L B, FITZGERALD N, LING J, et al. Matching the blanks: distributional similarity for relation learning[C]//Proceedings of the 57th annual meeting of the association for computational linguistics, Florence: Association for Computational Linguistics, 2019: 1-10.
[32] CHEN L, XU S, ZHU L, et al. A deep learning based method for extracting semantic information from patent documents[J]. Scientometrics, 2020, 125(1): 289-312.
Options
文章导航

/

〈 〉