摘要
针对传统协同过滤UserCF推荐算法,用户相似度计算数据维度单一、精确率低、用户兴趣动态漂移等问题,提出一种结合时间上下文和用户兴趣动态衰减模型的协同过滤推荐算法.在传统协同过滤算法中结合时间上下文优化用户相似度的计算,引入用户收敛兴趣衰减模型,动态评估用户兴趣变化.算法通过MovieLends-1M数据集进行实验验证,按照Fβ值、准确率、召回率与传统UserCF、UserCF-IIF算法进行对比分析.实验结果表明:本文算法精确率最高可达25.6%、召回率最高可达12.2%,相比UserCF、UserCF-IIF算法,在Fβ值方面本文算法具有明显提升.
For traditional collaborative filtering UserCF recommendation algorithm such as single dimension of data,low precision,dynamic drift of user interest,etc.,this paper presents a collaborative filtering recommendation algorithm which combines time context with dynamic attenuation model of user interest.In the traditional collaborative filtering algorithm,time context is combined to optimize the calculation of user similarity,and user convergence interest decay model is introduced to dynamically assess changes in user interest.The algorithm is validated by a comparative experiment on MovieLends-1M dataset and compared with traditional UserCF and UserCF-IIF algorithms in F_(β)value,precision and recall.The experimental results show that the improved recommendation model improves the recommendation performance compared with the previous model.
作者
王宁
崔德龙
宋丹
汤毅
WANG Ning;CUI Delong;SONG Dan;TANG Yi(College of Computer Science and Communications,Hunan Institute of Engineering,Xiangtan 411104,China)
出处
《湖南工程学院学报（自然科学版）》
2023年第1期40-46,共7页
Journal of Hunan Institute of Engineering(Natural Science Edition)
基金
教育部人文社会科学研究资助项目(20YJA880045)
湖南省普通高等学校教学改革项目(湘教通[2019]291号)
国家级大学生创新创业训练计划项目(S202111342016)。
关键词
时间上下文
动态兴趣
协同过滤
推荐系统
time context
dynamic interest
collaborative filtering
recommendation system
