摘要
针对协同过滤中存在的稀疏性问题,提出改进方法——BAS算法。该算法结合贝叶斯度量方法和奇异值降维分解方法,利用传统的基于奇异值分解获得活跃用户的邻居,通过改进后的相似性度量方法得出预测值。对改进后的算法进行理论分析和实验对比。结果表明,该方法在所用数据集上能够有效缓解数据稀疏性问题,并且改善推荐精度的准确性,在一定程度上提高了推荐引擎的推荐质量。
Targeting the sparsity problem of collaborative filtering ,an improved method- -BAS algorithms is proposed . The algorithm combines the Bayesian measure of dimensionality reduction and singular value decomposition method ,ob‐tained the neighbors of active users based on traditional singular value decomposition method ,to get the final predicted value w hich provided to users through improved similarity measure .Experimental results show that the method used in the data set can effectively alleviate the data sparseness problem ,and can improve the recommendation accuracy ,and the recommendation quality of engine to a certain extent .
出处
《软件导刊》
2015年第2期74-77,共4页
Software Guide
关键词
推荐引擎
协同过滤算法
数据稀疏
奇异值分解
Recommendation Engine Collaborative Filtering Algorithm Data Sparse Singular Value Decomposition
