基于改进相似度计算方法的协同过滤算法研究

协同过滤算法是目前推荐系统中应用最广泛的技术,相似度的计算是该算法中关键的一步,它直接影响到后续的目标用户邻居集的选取及评分预测,最终决定着推荐的准确度。在传统的基于用户的协同过滤中,相似度的计算未考虑用户评分差异和商品的热度对相似度计算的影响。论文引入平均评分修正因子和热门商品惩罚因子,对传统的相似度计算公式加以优化。实验表明,改进后的相似度算法在电影推荐时,平均绝对误差(MAE)值较其他相似度算法更低,有着更好的推荐效果。

Research on Collaborative Filtering Algorithm Based on Improved Similarity Calculation Method

Collaborative filtering algorithm is the most widely used techniques in the current recommendation system.The cal?culation of similarity is a key step in the algorithm.It directly affects the selection of subsequent target user neighbor sets and scoring prediction,which ultimately determines the accuracy of the recommendation.Intraditional user-based collaborative filtering,the calculation of similarity does not take into account the impact of user score differences and commodity heat on similarity calcula?tions.This paper introduces the average score correction factor and the popular commodity penalty factor to optimize the traditional similarity calculation formula.Experiments show that the improved similarity algorithm has lower Mean Absolute Error(MAE)value in film recommendation,and has better recommendation.