结合信任和相似度的随机游走推荐算法

A Random Walk Recommendation Algorithm Combining Trust and Similarity

摘要: 针对稀疏性和冷启动问题，提出一种结合信任和相似度的随机游走算法，利用两者的综合权重TS，应用于随机游走算法。实验结果表明，在全用户数据集和冷启动数据集中，算法比其他参照算法在准确率和覆盖率等方面均有提高，时间复杂度也有改善。本文的信任度采用数据集内用户评价的信任度，并没有采用信任度公式计算用户对其他用户的信任度。提出的算法改善了推荐精确度、覆盖率，优化了推荐质量。

Abstract: To address the similarity and cold start problem, a random walk algorithm combining trust and similarity is proposed with the weight TS. The experimental results indicate that the algorithm performs better with the all user data sets and cold start data sets than others in the aspect of accuracy rate, coverage rate as well as the time complexity. The trust value of the data set is used rather than the value computed by an effective method. The algorithm in this research improves the precision of recommendation, coverage rate and quality of recommendation.