# Do We Really Need to Drop Items with Missing Modalities in Multimodal Recommendation?

arXiv: https://arxiv.org/abs/2408.11767 · HTML: https://mlsanity.com/paper/2408.11767/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 21 Aug 2024.

## Built on by (most cited first, 7 of 7)

- [Disentangling and Generating Modalities for Recommendation in Missing Modality Scenarios](https://mlsanity.com/paper/2504.16352/) (2025)
- [NTSFormer: A Self-Teaching Graph Transformer for Multimodal Isolated Cold-Start Node Classification](https://mlsanity.com/paper/2507.04870/) (2025)
- [RGL: A Graph-Centric, Modular Framework for Efficient Retrieval-Augmented Generation on Graphs](https://mlsanity.com/paper/2503.19314/) (2025)
- [ICYM2I: The illusion of multimodal informativeness under missingness](https://mlsanity.com/paper/2505.16953/) (2025)
- [Benchmarking Multimodal Large Language Models for Missing Modality Completion in Product Catalogues](https://mlsanity.com/paper/2601.19750/) (2026)
- [MiMIC: Mitigating Visual Modality Collapse in Universal Multimodal Retrieval While Avoiding Semantic Misalignment](https://mlsanity.com/paper/2604.21326/) (2026)
- [Meta-Modal Agent: Sequential Evidence Routing for Missing-Modality Candidate Reranking](https://mlsanity.com/paper/2605.25007/) (2026)

## Abstract (arXiv)

Generally, items with missing modalities are dropped in multimodal recommendation. However, with this work, we question this procedure, highlighting that it would further damage the pipeline of any multimodal recommender system. First, we show that the lack of (some) modalities is, in fact, a widely-diffused phenomenon in multimodal recommendation. Second, we propose a pipeline that imputes missing multimodal features in recommendation by leveraging traditional imputation strategies in machine learning. Then, given the graph structure of the recommendation data, we also propose three more effective imputation solutions that leverage the item-item co-purchase graph and the multimodal similarities of co-interacted items. Our method can be plugged into any multimodal RSs in the literature working as an untrained pre-processing phase, showing (through extensive experiments) that any data pre-filtering is not only unnecessary but also harmful to the performance.