Applying random forest to drive recommendation

作者：Zhan, Le; Zhang, Jingwei; Yang, Qing^*; Lin, Yuming

来源：18th International Conference on Intelligent Data Engineering and Automated Learning, IDEAL 2017, 2017-10-30 To 2017-11-01.

DOI：10.1007/978-3-319-68935-7_51

• Attribute information
• Classification rules
• Online shopping
• Random forests
• Recommendation
• Recommendation methods
• Similarity computation
• User experience

Accurate information to users, which is required by online shopping, self-help travel etc. is very important to improve user experience. Recommendation is an important mechanism to match useful information to users with needs. Existing recommendation methods generally rely on massive similarity computation between users and recommended objects, which do not consider some fine-grained information and are not suitable for online recommendation. This paper introduces a novel model for recommendation, which merges classification strategy into recommendation and transforms classification rules into recommendation rules. Random forest is integrated with the proposed model for classification and then a ranking processing is carried out to find top-k users for recommendation. The proposed method makes full use of classification output and the relationships between users and recommended things, it is more suitable for online recommendation. Extensive experiments on different kinds of datasets indicate that the proposed method is effective.

• 出版日期2017