# HyperAgent4POI: Dynamic Semantic Message Passing on Multi-Agent Hypergraphs for Missing-Modality Recommendation

arXiv: https://arxiv.org/abs/2608.01846 · HTML: https://mlsanity.com/paper/2608.01846/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Agentic recommender systems, one of 16 directions in LLM agents and tool use.
- Builds on 19 of its 41 references that are in the map.
- First posted on arXiv 3 Aug 2026.

## Placed in

- LLM agents and tool use / [Agentic recommender systems](https://mlsanity.com/agents/agentic-recommender-systems/)

## Builds on (most cited first, 19 of 19)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [MiniLM: Deep Self-Attention Distillation for Task-Agnostic Compression of Pre-Trained Transformers](https://mlsanity.com/paper/2002.10957/) (2020)
- [Neural Graph Collaborative Filtering](https://mlsanity.com/paper/1905.08108/) (2019)
- [Self-supervised Graph Learning for Recommendation](https://mlsanity.com/paper/2010.10783/) (2020)
- [Are we really making much progress? Revisiting, benchmarking, and refining heterogeneous graph neural networks](https://mlsanity.com/paper/2112.14936/) (2021)
- [Representation Learning with Large Language Models for Recommendation](https://mlsanity.com/paper/2310.15950/) (2023)
- [On the Unreasonable Effectiveness of Feature propagation in Learning on Graphs with Missing Node Features](https://mlsanity.com/paper/2111.12128/) (2021)
- [MACRec: a Multi-Agent Collaboration Framework for Recommendation](https://mlsanity.com/paper/2402.15235/) (2024)
- [A Survey on LLM-powered Agents for Recommender Systems](https://mlsanity.com/paper/2502.10050/) (2025)
- [Graph Convolutional Networks for Graphs Containing Missing Features](https://mlsanity.com/paper/2007.04583/) (2020)
- [Learnable Expansion of Graph Operators for Multi-Modal Feature Fusion](https://mlsanity.com/paper/2410.01506/) (2024)
- [LLM-Guided Multi-View Hypergraph Learning for Human-Centric Explainable Recommendation](https://mlsanity.com/paper/2401.08217/) (2024)
- [A Multimodal Single-Branch Embedding Network for Recommendation in Cold-Start and Missing Modality Scenarios](https://mlsanity.com/paper/2409.17864/) (2024)
- [Generating with Fairness: A Modality-Diffused Counterfactual Framework for Incomplete Multimodal Recommendations](https://mlsanity.com/paper/2501.11916/) (2025)
- [LRMM: Learning to Recommend with Missing Modalities](https://mlsanity.com/paper/1808.06791/) (2018)
- [Meta-learning enhanced next POI recommendation by leveraging check-ins from auxiliary cities](https://mlsanity.com/paper/2308.09309/) (2023)
- [Disentangling and Generating Modalities for Recommendation in Missing Modality Scenarios](https://mlsanity.com/paper/2504.16352/) (2025)
- [Multi-Modal Hypergraph Enhanced LLM Learning for Recommendation](https://mlsanity.com/paper/2504.10541/) (2025)
- [Robust Multimodal Recommendation via Graph Retrieval-Enhanced Modality Completion](https://mlsanity.com/paper/2605.00670/) (2026)

## Abstract (arXiv)

Next Point-of-Interest (POI) recommendation benefits from textual and visual content that describes venue semantics, yet such content is often incomplete in real-world services. Missing modalities weaken POI representations and reduce the semantic evidence available for ranking. The resulting representations also provide unreliable evidence for modeling higher-order user--POI interactions. We propose HyperAgent4POI, which uses Dynamic Semantic Message Passing (DSMP) to perform modality completion and soft incidence refinement within each hypergraph layer. Persistent node agents share a frozen Llama backbone and use role-specific adapters to produce node-to-hyperedge messages. Semantic hyperedge motifs formed from these messages guide soft incidence scoring and modality completion. Final node representations are cached for online ranking without LLM calls. Experiments on three real-world LBSN datasets show consistent ranking gains over 15 baselines across modality-missing rates, while cached inference provides practical online efficiency. Under a 60% modality-missing rate, HyperAgent4POI improves NDCG@20 over the strongest baseline by 8.2% on average across the three datasets.