SemPOI-RL: Aligning LLM Semantic Reasoning for Interpretable Out-of-Town POI Sequential Generation
2026-08-31 • Computation and Language
Computation and LanguageArtificial Intelligence
AI summaryⓘ
The authors address the problem of generating travel destination sequences for users visiting new cities, which requires understanding users' travel preferences from their hometown trips and adapting these to different locations. They create a method called SemPOI-RL that fine-tunes a large language model to describe travel styles in natural language, making the preferences interpretable. Then, they use a special module to turn these styles into detailed travel plans while considering the order of visits. They also use reinforcement learning to improve the quality of recommendations. Their approach performs better than existing methods and clearly shows how travel styles influence planning.
large language modelssemantic reasoningpoint of interest (POI)sequence generationtravel recommendationreinforcement learningmasked autoencoderstyle-conditioned generationinterpretable machine learningcross-city adaptation
Authors
Yunqi Liu, Yang Zhang, Ruixing Zhang, Liangzhe Han, Yi Qiao, Tongyu Zhu, Leilei Sun
Abstract
Large language models (LLMs) exhibit strong semantic reasoning and open-ended generation abilities, but aligning these abilities with structured sequential generation remains challenging. This challenge is particularly evident in out-of-town (OOT) POI sequence generation, where a model must infer transferable travel intent from a user's hometown behaviors, adapt to cross-city interest drift, and generate a coherent destination trajectory under structural constraints. Existing approaches either rely on latent ID-based transfer with limited interpretability or directly use LLMs for sequence generation without explicitly grounding inferred semantics into position-aware predictions. To address this gap, we propose SemPOI-RL, a framework that aligns LLM semantic reasoning with structured sequence generation for interpretable OOT recommendation. Specifically, we first fine-tune an LLM to infer destination-oriented travel styles from users' hometown trajectories, using natural language as an interpretable semantic intermediate. We then introduce a Semantic POI Alignment Module (SPAM) to ground these inferred styles into a style-conditioned masked autoencoder for position-aware trajectory generation. Finally, we apply reinforcement learning with recommendation-oriented rewards to align LLM-generated styles with downstream sequence quality. Experiments on two real-world datasets show that SemPOI-RL consistently outperforms both traditional recommenders and direct LLM baselines, while providing interpretable style attribution across different phases of a trip. The code is available at https://github.com/Wind-Flipped/SemPOI-RL .