基于项目与用户的组合推荐算法研究

Research on the combined recommendation algorithm based on item and user

协同过滤技术是当前推荐系统中应用最广泛的推荐技术之一,随着系统用户规模的激增,传统的协同过滤技术存在实时性差、可扩展性差、数据稀疏性等问题。为了解决上述问题,提出了一种基于项目与用户的个性化组合推荐算法。首先,利用项目聚类对未评分项目进行评分预测,并填充用户-项目评分矩阵;再将项目聚类结果与用户行为特征相结合并对其进行用户聚类;最后,根据近邻相似性计算实现TOP-N推荐。实验表明,提出的组合推荐算法显著提高了推荐系统的准确性与实时性。

Collaborative filtering technology is one of the most widely used recommendation technologies in current recommendation system, with the fast growth in the amount of users,the traditional collaborative filtering technology has such problems as poor real-time performance,poor extendibility,data sparsity and so on. In order to solve the above problems,this paper presents a personalized combined recommendation algorithm based on item and user. First of all,it used project clustering to predict the score of the unrated items,and fill the user item rating matrix,then the project clustering results are combined with the user behavior characteristics to carry out user clustering,finally,according to the nearest neighbor similarity computation it achieved TOP-N recommendations. The experimental results show that the combined recommendation algorithm proposed in the paper improves the accuracy and real-time performance of the recommendation system significantly.