基于专利数据的电力标引信息挖掘技术研究  

Research on Power Indexing Information Mining Technology Based on Patent Data

作　　者：周长江 蔡榕 祝和明 王存超 郭晏 ZHOU Chang-jiang;CAI Rong;ZHU He-ming;WANG Cun-chao;GUO Yan(State Grid Jiangsu Electric Power Co.,Ltd.,Nanjing 210000 China)

机构地区：[1]国网江苏省电力有限公司,江苏南京210000

出　　处：《自动化技术与应用》2024年第4期118-121,共4页Techniques of Automation and Applications

基　　金：国网江苏省电力有限公司科技项目(J2019139)。

摘　　要：为了提高专利数据挖掘的准确性和可靠性,针对基于专利数据的电力标引信息挖掘技术进行研究。以SIPO专利数据库为数据源,生成专利数据序列。根据词嵌入模型设计Word2Vec,获取电力专利数据关联信息。根据数据关联融合结果,通过支持向量机分别训练相应子分类器,高效融合各子分类器,构建总分类模型完成分类决策,根据最终决策获取电力专利数据标引信息挖掘结果。实验结果表明,提出的挖掘与其他挖掘法相比查准率和查全率更高,具有可靠性。In order to improve the accuracy and reliability of patent data mining,research on power indexing information mining technology based on patent data is conducted.It uses the SIPO patent database as a data source to generate patent data sequences.It designs Word2Vec according to the word embedding model to obtain the related information of electric power patent data.According to the data association fusion results,the corresponding sub-classifiers are trained through support vector machines,and the sub-classifiers are efficiently fused to construct a general classification model to complete the classification decision.According to the final decision,the results of power patent data indexing information mining are obtained.Experimental results show that the proposed mining has higher precision and recall than other mining methods,and it is reliable.

关 键 词：专利数据 电力标引信息 关联融合 数据序列 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]