Attentive Neural Network Recommendation for E-Commerce Based on Various Implicit Feedback
Jinghua Zhu, Xu Guo, Shengchao Ma · 2018
With the Internet applications and information exploding exponentially, recommendation system has become an effective measure to solve information overloading. The traditional recommendation algorithm for users' implicit feedback is mainly based on collaborative filtering and Bayesian ranking methods, which are not fully utilize implicit feedback features such as click, collecting, adding to shopping cart and purchase. In order to make full use of various implicit feedback in the e-commence scenario and capture the dynamic temporal features of user-item interaction, this paper proposes a recommendation model which integrates use-embedding vector representation and self-attention mechanism. We conduct comparative experiments on public datasets from e-commerce websites. The evaluation results show that compared to the traditional recommendation models and other neural network models recently proposed, our model has better performance and generalization ability.