Tracking User Activities and Marketplace Dynamics in Classified Ads

Tracking users' posting activities in online classified ads and understanding the dynamics of their behavior is a topic of great importance with many implications. However, some of the underlying problems associated with modeling users and detecting their behavioral changes due to temporal and spatial variations have not been well-studied. In this paper, we develop a probabilistic model of user behavior based on the ads the user posts and the categories in which the ads are posted. The model can track some of the temporal changes in behavior, as revealed by our experiments on two classes of users monitored over a period of almost a year. We study the association between post categories and user groups, and show how temporal and seasonal changes can be detected. We further investigate a generative model for ad posts, based on user locations, and provide some evidence showing that the model is promising and that some interesting relationships can be identified.