20 August 2018 in IEEE Transactions on Knowledge and Data Engineering
Online interactive recommender systems strive to promptly suggest to consumers appropriate items according to the current context information. However, such context information is often unavailable in practice. Moreover, the lack of interaction records further worsens the recommendation performance. To address these issues, collaborative filtering (CF) and the online multi-armed bandit mechanisms are adopted in the online interactive recommendation settings, by assuming independent items (i.e., arms). Nonetheless, the assumption rarely holds in reality. In this paper, we study online interactive collaborative filtering problems by considering the dependencies among items. We explicitly formulate the item dependencies as the clusters on arms, where the arms within a single cluster share the similar latent topics. In light of the topic modeling techniques, we come up with a generative model to generate the items from their underlying topics. Furthermore, an efficient online algorithm based on particle learning is developed for inferring both latent parameters and states of our model by taking advantage of the fully adaptive inference strategy of particle learning. Additionally, our inferred model can be naturally integrated with existing multi-armed selection strategies in the online interactive collaborating setting. Empirical studies on two real-world applications demonstrate the effectiveness and efficiency of the proposed approach.
1 January 2021 in Information Sciences