Anchored Alignment: Preventing Positional Collapse in Multimodal Recommender Systems

TL;DR

AnchorRec introduces a novel multimodal recommendation framework that aligns modalities indirectly through anchors, preserving modality-specific structures and enhancing expressiveness without positional collapse, leading to competitive recommendation performance.

Contribution

It proposes AnchorRec, a lightweight, anchor-based alignment method that decouples alignment from representation learning in multimodal recommender systems.

Findings

Achieves competitive top N recommendation accuracy on Amazon datasets.

Improves multimodal expressiveness and coherence.

Effectively prevents positional collapse in multimodal embeddings.

Abstract

Multimodal recommender systems (MMRS) leverage images, text, and interaction signals to enrich item representations. However, recent alignment based MMRSs that enforce a unified embedding space often blur modality specific structures and exacerbate ID dominance. Therefore, we propose AnchorRec, a multimodal recommendation framework that performs indirect, anchor based alignment in a lightweight projection domain. By decoupling alignment from representation learning, AnchorRec preserves each modality's native structure while maintaining cross modal consistency and avoiding positional collapse. Experiments on four Amazon datasets show that AnchorRec achieves competitive top N recommendation accuracy, while qualitative analyses demonstrate improved multimodal expressiveness and coherence. The codebase of AnchorRec is available at https://github.com/hun9008/AnchorRec.