A Transformer-Based Fusion Recommendation Model For IPTV Applications
Heng Li, Hang Lei, Maolin Yang, Jinghong Zeng, Di Zhu, Shouwei Fu · 2020
Recommendation models are fundamental primitives of recommender systems that select contents to match users' interests. In real-life IPTV applications, a registered account is usually shared by multiple audiences, e.g., family members, invalidating the potential efforts in extracting user interests from the registered user profiles. Moreover, it is difficult to obtain explicit feedback from audiences due to the boring interactions between users and televisions in practice, i.e., using remote controls. Thus, designing recommendation algorithms to exploit user preference with implicit feedback is crucial for developing efficient recommender systems. In this paper, we present a deep model for IPTV applications, which generate recommendations using the implicit feedback of users. A fusion layer is designed on top of the Transformer framework to obtain the semantic preferences of audiences based on their behavior sequences. The presented model is shown to be effective in our real-life IPTV recommender systems. Empirical experiments on two open datasets also show that our method outperforms state of the art baselines.