Changing with Time: Modelling and Detecting User Lifecycle Periods in Online Community Platforms

In this paper we define the development of a user from entry to churn as his lifecycle that can be divided into discrete stages of development known as lifecycle periods. Prior work has examined how social networking site users have developed along isolated dimensions using lexical information [2] and social connections in the context of telecommunications networks [6]. We unify such dimensions by modelling and examining how users develop both socially and lexically, through contrasts of user properties (e.g. time-delimited in-degree distributions) against: (i) prior user properties; and (ii) the community in which the user is interacting. We identify salient traits of user development characterisable in the form of growth features, and demonstrate the applicability of such features within a vector space model by outperforming several baselines when detecting the lifecycle period of a given user.