Enhancing Large Language Models for Mobility Analytics with Semantic Location Tokenization.

Yile Chen, Yicheng Tao, Yue Jiang, Shuai Liu, Han Yu, Gao Cong.

We proposes QT-Mob, a framework that integrates semantically rich location tokenization with complementary fine-tuning objectives to enable LLMs to better understand mobility patterns and outperform prior models on tasks like next-location prediction and mobility recovery.