期刊文献+

融合隐性社交网络社团划分和协同过滤的推荐算法 被引量:1

A recommendation algorithm integrating implicit social network community division and collaborative filtering
下载PDF
导出
摘要 协同过滤算法是个性化推荐系统中广泛使用的经典算法。针对传统协同过滤算法存在的相似度计算不准确、可扩展性差等问题,设计了一种融合隐性社交网络社团划分和协同过滤的推荐算法ICDCF。该方法将用户对项目的共同兴趣视为社交关系。首先用考虑了用户隐性关系的改进的Jaccard相似系数衡量用户间的社交关系强弱,以用户为顶点、以用户相似度为连边的权值,构建无向加权的隐性社交网络;然后基于隐性社交网络,用谱聚类思想对用户进行社团划分;最后在社团内实施基于用户的协同过滤推荐。该方法可以避免协同过滤推荐阶段因共同评分项目少而导致的相似度计算不准确问题,同时可以减少搜索近邻的计算量,提高时间效率。在数据集MovieLens⁃100K和FilmTrust上的实验结果体现了ICDCF算法在推荐准确性和可扩展性方面的优势。 The collaborative filtering algorithm is a classical algorithm widely used in personalized recommendation systems.Since these algorithms face problems of inaccurate similarity calculation and poor scalability,this paper proposes a recommendation algorithm ICDCF integrating implicit social network community division and collaborative filtering.This algorithm regards the userscommon interests in item as social relations.Firstly,the improved Jaccard similarity coefficient considering the implicit relationship of users is used to measure the strength of the social relationship between users,and the undirected weighted implicit social network is constructed with the users as the nodes and the user similarity as the weight of the edges.And based on the implicit social network,users are divided into communities by spectral clustering.Then the user based collaborative filtering recommendation is implemented in the community.The algorithm can avoid inaccurate similarity calculation caused by the lack of common scoring items in the collaborative filtering recommendation stage,and reduce the cost of calculation in searching for neighbors in order to improve the time efficiency.The experimental results on the datasets of MovieLens⁃100k and FilmTrust show the advantages of ICDCF algorithm in recommendation accuracy and scalability.
作者 孙海岗 李玲娟 SUN Haigang;LI Lingjuan(School of Computer Science,Nanjing University of Posts and Telecommunications,Nanjing 210023,China)
出处 《南京邮电大学学报(自然科学版)》 北大核心 2023年第4期93-100,共8页 Journal of Nanjing University of Posts and Telecommunications:Natural Science Edition
基金 国家重点研发计划专项(2020YFB2104002) 江苏省重点研发计划(BE2019740)资助项目。
关键词 社团划分 Jaccard相似系数 谱聚类 协同过滤 推荐 community division Jaccard similarity coefficient spectral clustering collaborative filtering recommendation
  • 相关文献

参考文献10

二级参考文献175

共引文献442

同被引文献2

引证文献1

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部