nav emailalert searchbtn searchbox tablepage yinyongbenwen piczone journalimg journalInfo journalinfonormal searchdiv searchzone qikanlogo popupnotification paper paperNew
2025, 10, v.20 856-863
大模型检索之通用类结构检索的语义改写技巧
基金项目(Foundation):
邮箱(Email):
DOI:
摘要:

为了提高专利检索的效率质量以及专利审查质量,构建语义改写模型。介绍了基于模型的语义改写技巧,重点从技术主题应用领域的扩展、关键部件作用/功能的扩展和关键技术手段/技术特征权重值的增加这3个方向分析语义改写机理和改写方法,并结合深度学习的语义分析能力优化模型的语义改写能力。所提语义改写模型提高了大模型检索的命中率、召回率和精确率,可助力大模型在专利审查工作中的应用,为专利质量的提升和专利事业的可持续发展提供有力支持。

Abstract:

Semantic rewriting models were established to improve the efficiency and quality of patent retrieval, as well as the patent review quality, and model-based semantic rewriting skills were introduced. Mechanisms and methods of rewriting for semantic rewriting were analyzed in three directions with focus on expansion of technical subject application areas, expansion of key component roles/functions and increase of key technical means/technical feature weight values. Moreover, semantic rewriting capabilities of the model was optimized via semantic analysis capabilities combined with deep learning. The proposed rewriting model improves hit rate, recall rate and accuracy of large model retrieval, which facilitates the use of large models in patent review, and can provide strong support for improvements in patent quality and sustainable development of patent cause.

参考文献

[1]段然.报告显示:2023年全球专利申请量创历史新高[EB/OL].[2024-11-08]. http:∥www. cnipa. gov.cn/art/2024/11/8/art_15_384958. html.DUAN R. Report display:new height of history of global patent application inventory in 2023[EB/OL].[2024-11-08]. http:∥www. cnipa. gov. cn/art/2024/11/8/art_15_384958. html.(in Chinese)

[2]高慧,王铁军,范德增,等.从市场角度论高价值发明专利的价值回归[J].中国发明与专利,2023,20(12):72-80.GAO H, WANG T J, FAN D Z, et al. On the value regression of high value invertion patents from the market view[J]. China Invention&Patent, 2023,20(12):72-80.(in Chinese)

[3]张文,刘松元,刘梦瑶,等.大模型技术背景下美欧日韩各局最新审查动态[J].学术观察,2024(179):1-53.ZHAGN W, LIU S Y, LIU M Y, et al. Recent review dynamics in context of big model technology by office of the United States, Europe, Japan and Korea[J]. Academic Observer, 2024(179):1-53.(in Chinese)

[4]胡莹莹,李林.智能检索系统在交叉领域检索中的运用技巧[J].审查业务通讯,2024(10):74-81.HU Y Y, LI L. Exercise trick of intelligent retrieval system in cross-domain retrieval[J]. Review Service Communications, 2024(10):74-81.(in Chinese)

[5]方益奇,郑国威,崔航,等.便携式雷达三脚架系统结构设计[J].南方农机,2022, 53(14):153-155, 159.FANG Y Q, ZHENG G W, CUI H, et al. Structrual design of a portable rader tripod[J]. South Agricultural Machinery, 2022, 53(14):153-155, 159.(in Chinese)

[6]郭亮威,于洋.生活领域的机械类专利申请的检索策略[J].河南科技,2022, 41(14):143-146.GUO L W, YU Y. Search strategies for mechanical patent applications in the field of life[J]. Journal of Henan Science and Technology, 2022, 41(14):143-146.(in Chinese)

[7] AOURAGH S L, YOUSFI A, LAAROUSSI S,et al. A new estimate of the n-gram language model[J]. Procedia Computer Science, 2021, 189:211-215.

[8] SHEHAB A, AHMED R. Performance study on extractive text summarization using BERT models[J].Information, 2022, 13(2):67.

[9] DEV K, ASHRAF Z, MUHURI P K, et al. Deep autoencoder based domain adaptation for transfer learning[J]. Multimedia Tools and Applications, 2022,81(16):22379-22405.

基本信息:

中图分类号:G255.53;TP391.3;TP18

引用信息:

[1]冷林霞.大模型检索之通用类结构检索的语义改写技巧[J].中国科技论文,2025,20(10):856-863.

引用

GB/T 7714-2015 格式引文
MLA格式引文
APA格式引文
检 索 高级检索