ADS-POI: Agentic Spatiotemporal State Decomposition for Next Point-of-Interest Recommendation

arXiv cs.AI / 4/25/2026

💬 OpinionIdeas & Deep AnalysisTools & Practical UsageModels & Research

共有:

Key Points

The paper addresses next point-of-interest (POI) recommendation by modeling user mobility as spatiotemporal sequences and noting that existing approaches often compress history into a single latent state that entangles heterogeneous behaviors.
It proposes ADS-POI, which decomposes a user’s representation into multiple parallel evolving latent sub-states, each with its own spatiotemporal transition dynamics.
ADS-POI uses a context-conditioned selective aggregation mechanism to combine these sub-states into a decision state tailored to the current spatiotemporal context.
Experiments on three real-world Foursquare and Gowalla benchmark datasets show ADS-POI outperforms strong state-of-the-art baselines in full-ranking evaluations, improving both effectiveness and robustness for next-POI prediction.
The authors provide an open-source implementation of ADS-POI via the linked GitHub repository.

Abstract

Next point-of-interest (POI) recommendation requires modeling user mobility as a spatiotemporal sequence, where different behavioral factors may evolve at different temporal and spatial scales. Most existing methods compress a user's history into a single latent representation, which tends to entangle heterogeneous signals such as routine mobility patterns, short-term intent, and temporal regularities. This entanglement limits the flexibility of state evolution and reduces the model's ability to adapt to diverse decision contexts. We propose ADS-POI, a spatiotemporal state decomposition framework for next POI recommendation. ADS-POI represents a user with multiple parallel evolving latent sub-states, each governed by its own spatiotemporal transition dynamics. These sub-states are selectively aggregated through a context-conditioned mechanism to form the decision state used for prediction. This design enables different behavioral components to evolve at different rates while remaining coordinated under the current spatiotemporal context. Extensive experiments on three real-world benchmark datasets from Foursquare and Gowalla demonstrate that ADS-POI consistently outperforms strong state-of-the-art baselines under a full-ranking evaluation protocol. The results show that decomposing user behavior into multiple spatiotemporally aware states leads to more effective and robust next POI recommendation. Our code is available at https://github.com/YuZhenyuLindy/ADS-POI.git.