摘要
推荐系统可以为不同的用户定制个性化的网络服务,如何提供准确的推荐则成为其最大难点。针对传统推荐算法的稀疏性问题,提出基于用户行为特征的动态权重混合推荐算法。通过对数据集中的数据进行预处理,计算出不同用户对于不同物品的个性化行为特征指数,并将其引入相似度的计算中。依据用户评分数据稀疏性的差异计算出动态权重,并依此将基于用户内容的推荐和协同过滤推荐进行动态混合。实验结果表明,该算法在稀疏数据集中能有效降低推荐误差,提高推荐精度。
A recommendation system can personalize website service for different users, and how to provide accurate recommendations has become the biggest difficulty. Aiming at the sparsity problem of traditional recommendation algorithm, dynamic-weighted hybrid recommendation algorithm based on user behavior characteristics is proposed. Through the data preprocessing in dataset, the personalized behavior characteristic index of different users for different items is calculated and introduced into the similarity calculation. The dynamic weight is calculated according to the difference of the user’s rate data sparseness, and the user’s content recommendation and collaborative filtering recommendation are dynamically mixed. Experimental results show that the proposed algorithm can reduce the recommendation error effectively and improve the recommendation accuracy in the sparse data set.
出处
《计算机应用与软件》
2017年第4期316-321,共6页
Computer Applications and Software
关键词
行为特征
动态权重
混合推荐算法
User behavior characteristics
Dynamic-weighted
Hybrid recommendation algorithm
