Essay about A Unified Probabilistic Generative Model
i.e., users’ check-in records, are modeled as observed random variables, shown as shaded circles in
Figure 1. As a POI has both semantic and geographical attributes, we introduce two latent random variables, topic z and region r, which are responsible for generating them, respectively. Based on the two latent factors, TRM aims to model and infer users’ interests and spatial mobility patterns as well as their joint effect on users’ selection of POIs.
User Interest Modeling. Intuitively, a user chooses a POI by matching her personal in- terests with the contents of that POI. Inspired by the early work about user interest model- ing [Liu and Xiong 2013; Yin et al. 2014; Hu and Ester 2013; Hong et al. 2012], TRM adopts latent topics to characterize users’ interests to overcome the data sparsity of user-word matrix. Specifical- ly, we infer individual user’s interest distribution over a set of topics according to the contents (e.g., tags and categories) of her checked-in POIs, denoted as θu. Thus, the quality of topics is very important for accurately modeling users’ interests. To improve the…