Déjà vu: A Contextualized Temporal Attention Mechanism for Sequential Recommendation
This work addresses the challenge of improving recommendation accuracy for users in systems that rely on sequential behavior data, representing an incremental advancement by incorporating temporal and contextual factors into attention mechanisms.
The paper tackled the problem of sequential recommendation by addressing the neglect of temporal and context information in existing models, proposing a Contextualized Temporal Attention Mechanism that dynamically weighs historical actions based on when and how they occurred, and achieved consistent outperformance over state-of-the-art methods on two large public datasets.
Predicting users' preferences based on their sequential behaviors in history is challenging and crucial for modern recommender systems. Most existing sequential recommendation algorithms focus on transitional structure among the sequential actions, but largely ignore the temporal and context information, when modeling the influence of a historical event to current prediction. In this paper, we argue that the influence from the past events on a user's current action should vary over the course of time and under different context. Thus, we propose a Contextualized Temporal Attention Mechanism that learns to weigh historical actions' influence on not only what action it is, but also when and how the action took place. More specifically, to dynamically calibrate the relative input dependence from the self-attention mechanism, we deploy multiple parameterized kernel functions to learn various temporal dynamics, and then use the context information to determine which of these reweighing kernels to follow for each input. In empirical evaluations on two large public recommendation datasets, our model consistently outperformed an extensive set of state-of-the-art sequential recommendation methods.