INTRODUCTION Multiple data source mining is the process of identifying potentially useful patterns from different data sources, or datasets (Zhang et al., 2003). Group pattern discovery systems for mining different data sources are based on local pattern-analysis strategy, mainly including logical systems for information enhancing, a pattern discovery system, and a post-pattern-analysis system.

Discovering patterns and detecting anomalies in individual travel behavior is a crucial problem both research practice. In this paper, we address by building probabilistic framework to model spatiotemporal data (e.g., trip records trajectory data). We develop two-dimensional latent Dirichlet allocation (LDA) characterize the generative mechanism of each traveler. This introduces two separate fa...