基于多方面评分的景点协同推荐算法

Collaborative recommendation for scenic spots based on multi-aspect ratings

针对传统的协同过滤(collaborative filtering,CF)推荐模型中利用单一的总体评分进行相似性计算,但总体评分不能准确反映用户对物品喜好的问题,提出基于多方面评分的景点协同推荐算法。该算法综合利用用户对景点在景色、趣味性、性价比三个方面的评分计算用户或景点之间的相似性,进而计算目标用户对目标景点的总体评分。试验结果表明:在相似性计算中引入景点在这三个方面的评分信息后,推荐结果的均方根误差、平均绝对误差、覆盖率、准确率和F-度量指标都得到了改善。

The simplex overall ratings are used to compute the similarities between users and items in the model of traditional collaborative filtering recommendation, but it can＇t correctly depict the users＇ true preferences. In order to solve this problem, a collaborative scenic spots recommendation algorithm based on multi-aspect ratings was proposed, which integrated the ratings of the scenery, interesting and cost performance of spots to compute the similarities to predict the overall ratings of an active user for a target spot. Experimental results showed that, after introducing the information of multi-aspect ratings, the proposed method improved the accuracy of prediction score, coverage and F-measure and re- duced the predicting error of root-mean-square and mean-absolute.