摘要
[目的/意义]提出一种融合专利属性信息和用户行为分析的个性化专利推荐方法,以更好地满足用户的专利文献需求。[方法/过程]首先量化用户对专利文献的检索与操作行为,计算用户之间的相似性指数;然后结合专利文献的属性信息(如IPC分类号、发明人等),以实现个性化推荐。将该方法应用于自研的专利推荐系统中,并与其他协同过滤推荐方法进行对比。[局限]可能受到数据集的特定性限制,如依赖于特定的用户行为和专利文献数据集。此外,实验结果可能会因系统中的用户和专利文献选择不同而有所偏差。[结果/结论]实验结果表明,该个性化推荐方法优于传统的用户协同过滤推荐方法,能够更好地理解用户的需求和偏好,从而提高推荐的准确性与效率。
[Objective/Significance]This paper aims to propose a personalized patent recommendation method by integrating patent attribute information and user behavior analysis to better meet users’needs for patent literature.[Methods/Processes]The study first quantifies users’search and operation behaviors towards patent literature and calculates the similarity index between users.Then,it incorporates the attribute information of patent literature(e.g.,IPC classification,inventors)to implement personalized recommendations.This method is applied to a self-developed patent recommendation system and compared with other collaborative filtering recommendation methods.[Limitations]The research may be limited by the specificity of the data,such as the reliance on specific user behaviors and patent literature datasets.Moreover,the experimental results may vary depending on the selection of users and patent literature in the system.[Results/Conclusions]The experimental results show that this personalized recommendation method outperforms the traditional collaborative filtering recommendation method.It better understands users’needs and preferences,thereby improving the accuracy and efficiency of recommendations.
作者
赵学铭
王刚
ZHAO Xueming;WANG Gang(Science and Technology Office of Tianjin Medical University,Tianjin 300070,China;Library of Tianjin Medical University,Tianjin 300070,China)
出处
《情报工程》
2025年第1期42-53,共12页
Technology Intelligence Engineering
基金
2023年度天津市专利转化专项计划专项基金项目“供给侧专利培育与转化促进项目研究”。
关键词
用户行为分析
协同过滤
个性化推荐
专利服务
User Behavior Analysis
Collaborative Filtering
Personalized Recommendation
Patent Services
