Start Date

14-12-2012 12:00 AM

Description

The success of social media websites—Web 2.0 platforms that enable users to communicate among themselves—hinges on users’ ability to generate content and respond to content supplied by others. Thus, understanding the current and expected patterns of participation among their users is a fundamental concern for the managers of social media websites. However, current prevailing prediction approaches use very simple benchmark models. We offer a novel approach for estimating the number of active users and predicting future participation. Using a unique data set from online forums, we demonstrate how probability models (specifically, the geometric/beta-Bernoulli model) traditionally used for customer loyalty analysis can be successfully used to assess participation patterns, and thus the future value of online communities. We further explore the factors that affect the usefulness of this approach. Compared with current methods, our approach generates better estimations of future visits as well as future contribution levels.

Share

COinS
 
Dec 14th, 12:00 AM

Predicting Participation in Social Media Sites by Analyzing User Participation Patterns

The success of social media websites—Web 2.0 platforms that enable users to communicate among themselves—hinges on users’ ability to generate content and respond to content supplied by others. Thus, understanding the current and expected patterns of participation among their users is a fundamental concern for the managers of social media websites. However, current prevailing prediction approaches use very simple benchmark models. We offer a novel approach for estimating the number of active users and predicting future participation. Using a unique data set from online forums, we demonstrate how probability models (specifically, the geometric/beta-Bernoulli model) traditionally used for customer loyalty analysis can be successfully used to assess participation patterns, and thus the future value of online communities. We further explore the factors that affect the usefulness of this approach. Compared with current methods, our approach generates better estimations of future visits as well as future contribution levels.