UeLP: accurate user linkage across social platforms against location errors.

The article focuses on UeLP, a novel method for accurate cross-platform user linkage across instant messaging social platforms that addresses the challenges posed by location data errors and sparsity. UeLP employs a multi-grained spatio-temporal grid to organize user trajectory data, extracts user features through a spatio-temporal weight model combining grid contribution and motion sequence similarity, and filters candidate trajectories based on coarse-grained movement states to improve efficiency. Experiments on six public datasets and one self-built dataset demonstrate that UeLP outperforms four existing location-based methods, achieving average improvements of 15–45% in top-1 accuracy and 13–35% in top-3 accuracy. The study highlights the method's robustness against noisy and sparse location data typical of instant messaging platforms and suggests future work incorporating multimodal data for enhanced cross-platform user linkage.