Attribute-driven Disentangled Representation Learning for Multimodal Recommendation

Zhenyang Li, Fan Liu, Yinwei Wei, Zhiyong Cheng, Liqiang Nie, Mohan Kankanhalli

Abstract: Recommendation algorithms forecast user preferences by correlating user and item representations derived from historical interaction patterns. In pursuit of enhanced performance, many methods focus on learning robust and independent representations by disentangling the intricate factors within interaction data across various modalities in an unsupervised manner. However, such an approach obfuscates the discernment of how specific factors (e.g., category or brand) influence the outcomes, making it challenging to regulate their effects. In response to this challenge, we introduce a novel method called Attribute-Driven Disentangled Representation Learning (short for AD-DRL), which explicitly incorporates attributes from different modalities into the disentangled representation learning process. By assigning a specific attribute to each factor in multimodal features, AD-DRL can disentangle the factors at both attribute and attribute-value levels. To obtain robust and independent representations for each factor associated with a specific attribute, we first disentangle the representations of features both within and across different modalities. Moreover, we further enhance the robustness of the representations by fusing the multimodal features of the same factor. Empirical evaluations conducted on three public real-world datasets substantiate the effectiveness of AD-DRL, as well as its interpretability and controllability.

Primary Subject Area: [Engagement] Multimedia Search and Recommendation

Secondary Subject Area: [Engagement] Multimedia Search and Recommendation

Relevance To Conference: This work introduces Attribute-Driven Disentangled Representation Learning (AD-DRL), a novel method that significantly advances multimedia/multimodal processing by enhancing the independence and robustness of data representations. AD-DRL specifically targets the complex and intertwined factors within multimodal data by disentangling these factors at both attribute and attribute-value levels. This disentanglement allows for a clearer separation and identification of the various influences on user preferences, crucial for processing heterogeneous data from different modalities. Utilizing the multimodal representations obtained from AD-DRL can greatly enhance the interpretability and controllability of the recommendation results. Besides improving the performance of recommender systems, AD-DRL also provides an approach for more nuanced analysis and utilization of complex multimodal datasets.

Submission Number: 2767