【摘要】 针对传统的协调过滤推荐算法利用单一评分矩阵带来的数据稀疏性问题,提出一种基于多维特征聚类和用户评分的景点推荐算法。划分用户类别,使用基于属性权重的加权K-means聚类算法将表示用户特征的多维指标数值进行聚类;确定目标用户类别,引入用户的推荐可信度和质量可信度并形成评分可信度,将评分可信度和评分相似度结合平衡因子来计算用户之间的相似度,优化传统的相似度推荐算法。实验结果表明,该算法降低了数据的稀疏性,提高了推荐精度,具有更好的稳定性。更多还原

【Abstract】 To solve the problem of data sparsity of traditional coordinated filtering recommendation algorithm by using single score matrix,scenic spots recommendation algorithm based on the multi-dimensional feature clustering and user score was proposed.The multidimensional index values of user characteristics were clustered using the k-means algorithm based on optimization of attribute weight.Target user’s category was determined.The score reliability was designed by recommendation credibility and quality credibility and it was used to combine with rating similarity to compute the similarities between users.The similarity algorithm was optimized.Experimental results show that this algorithm not only reduces the data sparsity,but also improves the recommended precision with better stability.更多还原