基于辅助信息特征融合的序列推荐算法OA

The sequential recommendation predicts the user's future action based on the user's historical behavior.In order to improve the accuracy of next-item prediction,the modeling information includes the product category and other auxiliary informa-tion.The inner correlation of auxiliary information is ignored because the current algorithm fuses the sequential information and aux-iliary information earlier.In response to the above problems,this study proposes a Auxiliary Information Feature Fusion for Sequen-tial Recommendation(AIFF),which uses the attention mechanism to extract the inherent multi-granularity relationship of auxiliary information.The fused auxiliary information representation is used as the attention weight.And the sequential information is used as the value of attention.This paper inputs it into attention mechanism and neural network to fuse sequence information to make recom-mendations.The contrast experiment on three recommendation datasets shows AIFF has achieved good experimental results.At the same time,the universality experiment shows that the auxiliary information feature fusion solution can be incorporated into other at-tention sequential recommendations.