CoCoB: Adaptive Collaborative Combinatorial Bandits for Online Recommendation

TL;DR

CoCoB introduces a two-sided adaptive collaborative combinatorial bandit framework that combines a Bayesian user-bandit for adaptive neighbor discovery with a linear contextual item-bandit for top-$K$ recommendations. By using a similarity posterior with a threshold $\gamma$, CoCoB dynamically leverages neighbors when informative and relies on the target user otherwise, while maintaining computational efficiency. Theoretical regret analysis under a linear contextual setting is provided, and empirical evaluation on three real-world datasets shows CoCoB achieving a notable improvement, including an average 2.4% gain in F1 over state-of-the-art baselines. This approach offers a practical and scalable path to incorporating collaborative signals into online recommendations, with robust performance across diverse datasets and parameter settings.

Abstract

Clustering bandits have gained significant attention in recommender systems by leveraging collaborative information from neighboring users to better capture target user preferences. However, these methods often lack a clear definition of similar users and face challenges when users with unique preferences lack appropriate neighbors. In such cases, relying on divergent preferences of misidentified neighbors can degrade recommendation quality. To address these limitations, this paper proposes an adaptive Collaborative Combinatorial Bandits algorithm (CoCoB). CoCoB employs an innovative two-sided bandit architecture, applying bandit principles to both the user and item sides. The user-bandit employs an enhanced Bayesian model to explore user similarity, identifying neighbors based on a similarity probability threshold. The item-bandit treats items as arms, generating diverse recommendations informed by the user-bandit's output. CoCoB dynamically adapts, leveraging neighbor preferences when available or focusing solely on the target user otherwise. Regret analysis under a linear contextual bandit setting and experiments on three real-world datasets demonstrate CoCoB's effectiveness, achieving an average 2.4% improvement in F1 score over state-of-the-art methods.

Figures (6)

• Figure 1: A personalized recommendation task modeled as a clustering bandit problem. Black labels denote recommendation concepts, while blue labels represent bandit concepts. The Environment box determines the target user's cluster.
• Figure 2: The general framework of the proposed CoCoB recommendation.
• Figure 3: Recommendation performance of different algorithms on IJCAI-15.
• Figure 4: Recommendation performance of different algorithms on Retailrocket.
• Figure 5: Recommendation performance of different algorithms on Yoochoose.