Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Unifying Inductive, Cross-Domain, and Multimodal Learning for Robust and Generalizable Recommendation

Chanyoung Chung, Kyeongryul Lee, Sunbin Park, Joyce Jiyoung Whang

TL;DR

MICRec advances recommender systems by unifying inductive, multimodal, and cross-domain learning to handle unseen users/items and cross-domain data. It introduces template-driven inductive representations, modality-based aggregation using text and image encodings, and a cross-domain contrastive objective to align domain representations via overlapping users. Experiments on three Amazon-based cross-domain datasets show MICRec achieving substantial gains over INMO and other baselines, particularly in data-scarce settings. The approach offers practical impact for real-world systems operating across diverse domains with multimodal content and continual entity emergence, and suggests future directions with expansive language and knowledge-graph integration.

Abstract

Recommender systems have long been built upon the modeling of interactions between users and items, while recent studies have sought to broaden this paradigm by generalizing to new users and items, incorporating diverse information sources, and transferring knowledge across domains. Nevertheless, these efforts have largely focused on individual aspects, hindering their ability to tackle the complex recommendation scenarios that arise in daily consumptions across diverse domains. In this paper, we present MICRec, a unified framework that fuses inductive modeling, multimodal guidance, and cross-domain transfer to capture user contexts and latent preferences in heterogeneous and incomplete real-world data. Moving beyond the inductive backbone of INMO, our model refines expressive representations through modality-based aggregation and alleviates data sparsity by leveraging overlapping users as anchors across domains, thereby enabling robust and generalizable recommendation. Experiments show that MICRec outperforms 12 baselines, with notable gains in domains with limited training data.

Unifying Inductive, Cross-Domain, and Multimodal Learning for Robust and Generalizable Recommendation

TL;DR

MICRec advances recommender systems by unifying inductive, multimodal, and cross-domain learning to handle unseen users/items and cross-domain data. It introduces template-driven inductive representations, modality-based aggregation using text and image encodings, and a cross-domain contrastive objective to align domain representations via overlapping users. Experiments on three Amazon-based cross-domain datasets show MICRec achieving substantial gains over INMO and other baselines, particularly in data-scarce settings. The approach offers practical impact for real-world systems operating across diverse domains with multimodal content and continual entity emergence, and suggests future directions with expansive language and knowledge-graph integration.

Abstract

Recommender systems have long been built upon the modeling of interactions between users and items, while recent studies have sought to broaden this paradigm by generalizing to new users and items, incorporating diverse information sources, and transferring knowledge across domains. Nevertheless, these efforts have largely focused on individual aspects, hindering their ability to tackle the complex recommendation scenarios that arise in daily consumptions across diverse domains. In this paper, we present MICRec, a unified framework that fuses inductive modeling, multimodal guidance, and cross-domain transfer to capture user contexts and latent preferences in heterogeneous and incomplete real-world data. Moving beyond the inductive backbone of INMO, our model refines expressive representations through modality-based aggregation and alleviates data sparsity by leveraging overlapping users as anchors across domains, thereby enabling robust and generalizable recommendation. Experiments show that MICRec outperforms 12 baselines, with notable gains in domains with limited training data.
Paper Structure (13 sections, 8 equations, 3 figures, 4 tables)

This paper contains 13 sections, 8 equations, 3 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Problem Statement
  4. Methodology
  5. Template-Driven Inductive Modeling
  6. Modality-Based Aggregation
  7. Cross-Domain Contrastive Learning
  8. Experiments
  9. Experimental Details
  10. Quantitative Performance Comparison
  11. Effectiveness of MICRec over INMO
  12. Qualitative Study of Multimodal Similarity
  13. Conclusion and Future Work

Figures (3)

  • Figure 1: Overview of MICRec. MICRec enables inductive, multimodal, and cross-domain recommendation based on template-driven inductive modeling, modality-based aggregation, and contrastive learning on overlapping users between two domains.
  • Figure 2: Comparison of MICRec and INMO on NDCG@$N$ across Varying Number of Recommended Items ($N$). MICRec consistently outperforms INMO across all domains.
  • Figure 3: Case Study of User Similarity. The images correspond to the products listed in the purchase histories. In all domains, target users and their most similar users exhibit consistent purchase patterns, with overlapping items highlighted in purple.