Application of an Asset Bubble Model to Microblog Data Analytics

As microblogs such as Twitter become commonplace, greater importance needs to be given to data generated in those media as sources for information mining. This paper proposes a computationally simple model for future outcome prediction at current time using shot-term data generated from microblogs (tweets in particular). The model adopts an asset price bubble model, which is non-probabilistic. A threshold approach is used for predictions. A method to generate time series data from tweets is also presented. The methodology is illustrated using Twitter data. It is simple and easy to implement.