Detecting Present Events to Predict Future: Detection and Evolution of Events on Twitter

Event is not a one-off occurrence of something, rather it is a continuous chain of small events. Along with event detection, the event evolution is equally important. Most existing methods ignore the evolution of the events and further fails to identify the influential spreaders of that event. Moreover, for the impact of the continuous developing events, predicting the linkages with other premature events, will help in the domains of economy such as commodities and stock markets. Twitter is widely used as an effective source of data collection and provides unique keywords (hashtags). However, it does not provide insights about the trend which makes it difficult to detect events. The research experiment environment and preliminary results presented in this paper are currently based upon the Brexit’s historical dataset. However, we plan to define proper way to simultaneously snapshot whole dynamic dataset at different times and eventually use real-time data. Our motivation for the research is to develop a new data analytic system and supporting techniques to find events for enhancing the decision-making,,,, process.