基于深度学习与多分类轮询机制的高质量“卡脖子” 技术专利识别模型——以专利申请文件为研究主体

Xuefeng Zhao; Delin Wu; Weiwei Wu; Zhuoluo Sun; Jinjin Hu; Ying Lian; Jiayu Shan

doi:10.11925/infotech.2096-3467.2022.0721

基于深度学习与多分类轮询机制的高质量“卡脖子” 技术专利识别模型——以专利申请文件为研究主体

Translated title of the contribution: Identifying High-Quality Technology Patents Based on Deep Learning and Multi-Category Polling Mechanism——Case Study of Patent Applications

Xuefeng Zhao
, Delin Wu
, Weiwei Wu^*
, Zhuoluo Sun
, Jinjin Hu
, Ying Lian
, Jiayu Shan

^*Corresponding author for this work

School of Management, Harbin Institute of Technology
Shenzhen Ward Intellectual Property Agency
Ltd.

Research output: Contribution to journal › Article › peer-review

8 Scopus citations

Abstract

[Objective] This paper addresses the issues of the traditional single classification method, which cannot effectively identify high-quality“bottleneck”technology patents. [Methods] We developed a multi-category polling model (LSTM-Seq-BERT) with LSTM, Word2Vec, and BERT to identify high-quality“bottleneck” patents from the application documents. Moreover, we constructed a corresponding multi-level label system for the model with IPC number as the primary classification labels and authorization status as the secondary classification labels. [Results] The accuracy of identifying high-quality“bottleneck”technology patents was increased to 88. 1%. [Limitations] We only utilized patents from the Hongkong-Macau-Guangdong Greater Bay Area, resulting in data imbalance. [Conclusions] The proposed model can enhance the accuracy of identifying high-quality“bottleneck”technology patents and possesses practical value.

Translated title of the contribution	Identifying High-Quality Technology Patents Based on Deep Learning and Multi-Category Polling Mechanism——Case Study of Patent Applications
Original language	Chinese (Traditional)
Pages (from-to)	30-45
Number of pages	16
Journal	Data Analysis and Knowledge Discovery
Volume	7
Issue number	8
DOIs	https://doi.org/10.11925/infotech.2096-3467.2022.0721
State	Published - Aug 2023
Externally published	Yes

Access to Document

10.11925/infotech.2096-3467.2022.0721

Cite this

@article{9be4fcb6b0654860aeb7947ceab9e61a,

title = "基于深度学习与多分类轮询机制的高质量{\textquotedblleft}卡脖子{\textquotedblright} 技术专利识别模型{\textemdash}{\textemdash}以专利申请文件为研究主体",

abstract = "[Objective] This paper addresses the issues of the traditional single classification method, which cannot effectively identify high-quality{\textquotedblleft}bottleneck{\textquotedblright}technology patents. [Methods] We developed a multi-category polling model (LSTM-Seq-BERT) with LSTM, Word2Vec, and BERT to identify high-quality{\textquotedblleft}bottleneck{\textquotedblright} patents from the application documents. Moreover, we constructed a corresponding multi-level label system for the model with IPC number as the primary classification labels and authorization status as the secondary classification labels. [Results] The accuracy of identifying high-quality{\textquotedblleft}bottleneck{\textquotedblright}technology patents was increased to 88. 1\%. [Limitations] We only utilized patents from the Hongkong-Macau-Guangdong Greater Bay Area, resulting in data imbalance. [Conclusions] The proposed model can enhance the accuracy of identifying high-quality{\textquotedblleft}bottleneck{\textquotedblright}technology patents and possesses practical value.",

keywords = "Application Document, BERT, Bottleneck{\textquotedblright}Technology, LSTM, Multi-Category Polling Patent",

author = "Xuefeng Zhao and Delin Wu and Weiwei Wu and Zhuoluo Sun and Jinjin Hu and Ying Lian and Jiayu Shan",

note = "Publisher Copyright: {\textcopyright} 2023 The Author(s).",

year = "2023",

month = aug,

doi = "10.11925/infotech.2096-3467.2022.0721",

language = "繁体中文",

volume = "7",

pages = "30--45",

journal = "Data Analysis and Knowledge Discovery",

issn = "2096-3467",

publisher = "Chinese Academy of Sciences",

number = "8",

}

TY - JOUR

T1 - 基于深度学习与多分类轮询机制的高质量“卡脖子” 技术专利识别模型——以专利申请文件为研究主体

AU - Zhao, Xuefeng

AU - Wu, Delin

AU - Wu, Weiwei

AU - Sun, Zhuoluo

AU - Hu, Jinjin

AU - Lian, Ying

AU - Shan, Jiayu

PY - 2023/8

Y1 - 2023/8

N2 - [Objective] This paper addresses the issues of the traditional single classification method, which cannot effectively identify high-quality“bottleneck”technology patents. [Methods] We developed a multi-category polling model (LSTM-Seq-BERT) with LSTM, Word2Vec, and BERT to identify high-quality“bottleneck” patents from the application documents. Moreover, we constructed a corresponding multi-level label system for the model with IPC number as the primary classification labels and authorization status as the secondary classification labels. [Results] The accuracy of identifying high-quality“bottleneck”technology patents was increased to 88. 1%. [Limitations] We only utilized patents from the Hongkong-Macau-Guangdong Greater Bay Area, resulting in data imbalance. [Conclusions] The proposed model can enhance the accuracy of identifying high-quality“bottleneck”technology patents and possesses practical value.

AB - [Objective] This paper addresses the issues of the traditional single classification method, which cannot effectively identify high-quality“bottleneck”technology patents. [Methods] We developed a multi-category polling model (LSTM-Seq-BERT) with LSTM, Word2Vec, and BERT to identify high-quality“bottleneck” patents from the application documents. Moreover, we constructed a corresponding multi-level label system for the model with IPC number as the primary classification labels and authorization status as the secondary classification labels. [Results] The accuracy of identifying high-quality“bottleneck”technology patents was increased to 88. 1%. [Limitations] We only utilized patents from the Hongkong-Macau-Guangdong Greater Bay Area, resulting in data imbalance. [Conclusions] The proposed model can enhance the accuracy of identifying high-quality“bottleneck”technology patents and possesses practical value.

KW - Application Document

KW - BERT

KW - Bottleneck”Technology

KW - LSTM

KW - Multi-Category Polling Patent

UR - https://www.scopus.com/pages/publications/85176561687

U2 - 10.11925/infotech.2096-3467.2022.0721

DO - 10.11925/infotech.2096-3467.2022.0721

M3 - 文章

AN - SCOPUS:85176561687

SN - 2096-3467

VL - 7

SP - 30

EP - 45

JO - Data Analysis and Knowledge Discovery

JF - Data Analysis and Knowledge Discovery

IS - 8

ER -

基于深度学习与多分类轮询机制的高质量“卡脖子” 技术专利识别模型——以专利申请文件为研究主体

Abstract

Access to Document

Other files and links

Fingerprint

Cite this