摘要
针对推荐算法数据稀疏及聚类中心点敏感问题,提出了一种基于用户偏好和麻雀搜索聚类的协同过滤推荐算法。首先使用评分偏好模型对原用户项目矩阵进行修正,得到新的用户偏好-项目矩阵。利用麻雀搜索对聚类中心点进行优化,从目标用户所在簇内得到最近邻,提高了算法迭代速度,改善了聚类中心点敏感的问题。使用相似度公式对目标用户未评分项目进行预测,并完成推荐。实验结果表明,相较于其他几种推荐算法,准确度提高了4到6个百分点。
To solve the problem of sparse data of recommendation algorithm and sensitive clustering center point,a recommendation algorithm based on user preference and sparrow search clustering collaborative filtering is proposed.The original user item matrix is modified by using the rating preference model,and the new user preference item matrix is obtained.The sparrow search is used to optimize the clustering center point,and the nearest neighbor is obtained from the target user's cluster.It improves the iteration speed of the algorithm and improves the sensitivity of the clustering center point.The similarity formula is used to predict the unrated items of target users,and the recommendation is completed.Experimental results show that the accuracy is improved by 4 to 6 percentage points compared with other recommendation algorithms.
作者
聂晓明
高鹏翔
NIE Xiao-ming;GAO Peng-xiang(School of Data Science and Software Engineering, Qingdao University, Qingdao 266071, China)
出处
《青岛大学学报（自然科学版）》
CAS
2021年第1期70-76,共7页
Journal of Qingdao University(Natural Science Edition)
基金
山东省自然科学基金(批准号:ZR2019PEE018)资助。
关键词
评分偏好
麻雀搜索
协同过滤
推荐精度
scoring preference
sparrow search
collaborative filtering
recommendation accuracy
