Rethinking Multi-Interest Learning for Candidate Matching in Recommender Systems

Abstract

Existing research efforts for multi-interest candidate matching in recommender systems mainly focus on improving model architecture or incorporating additional information, neglecting the importance of training schemes. This work revisits the training framework and uncovers two major problems hindering the expressiveness of learned multi-interest representations. First, the current training objective (i.e., uniformly sampled softmax) fails to effectively train discriminative representations in a multi-interest learning scenario due to the severe increase in easy negative samples. Second, a routing collapse problem is observed where each learned interest may collapse to express information only from a single item, resulting in information loss. To address these issues, we propose the REMI framework, consisting of an Interest-aware Hard Negative mining strategy (IHN) and a Routing Regularization (RR) method. IHN emphasizes interest-aware hard negatives by proposing an ideal sampling distribution and developing a Monte-Carlo strategy for efficient approximation. RR prevents routing collapse by introducing a novel regularization term on the item-to-interest routing matrices. These two components enhance the learned multi-interest representations from both the optimization objective and the composition information. REMI is a general framework that can be readily applied to various existing multi-interest candidate matching methods. Experiments on three real-world datasets show our method can significantly improve state-of-the-art methods with easy implementation and negligible computational overhead. The source code will be released. CCS CONCEPTS• Information systems → Recommender systems.

Is Contrastive Learning Necessary? A Study of Data Augmentation vs Contrastive Learning in Sequential Recommendation

“…Except for the aforementioned models, attention-based models have also been intensively studied and widely adopted in sequential recommendation tasks [19,37,55]. Besides, there are many interesting ongoing works focusing on other techniques like contrastive learning [4,27,48,63,64], reinforcement learning [51], multi-interest learning [49], large language model [25,26,62] and relation awareness [14].…”

A Survey of GPT-3 Family Large Language Models Including ChatGPT and GPT-4

“…For example, popular platforms like YouTube and Netflix use recommendation systems to suggest relevant videos and platforms like Amazon use recommendation systems to suggest relevant products to the user [237]. The commonly used approaches for recommendation systems are based on collaborative filtering [238], content-based [239] and knowledge-based [240]. The performance of traditional recommendation systems is limited by a number of issues like cold-start problem, poor generalization across domains and lack of explainability [241], [242].…”

The Distribution of Etiology and Clinical Presentations of Birth Defects: A Preliminary Assessment of China Birth Cohort Study

“…In early ages, collaborative filtering-based methods [5,6,44,62] are primarily used to model the user's behavior patterns from the user-item interactions. Later on, with the introduction of user and item side information into recommendation systems, content-based recommendation [36,37,40,53,58] and knowledge-based recommendation [2,8,16,18] have gained attention due to their ability to provide personalized recommendations.…”