Real Time Semantic Events Detection from Social Media Stream

Real time monitoring of twitter tweet streams for events has popularity in the last decade. This provides effective information for government, business and other organization to know what happening right now. The task comprises many challenges including the processing of large volume of data in real time and high levels of noise. The main objective of this work is timely detection of semantic bursty events which have happened recently and discovery of their evolutionary patterns along the timeline. We present semantic burst detection in adaptive time windows and then retrieve evolutionary patterns of burst over time period. Burst is the task of finding unexpected change of some quantity in real time tweet stream. Moreover burst is highly depending on the sampled time window size and threshold values. Thus we propose how to adjust time windows sizes and threshold values for burst detection in real time. To get accurate burst from real time twitter stream, semantic words and phrase extraction from noise polluted text stream is proposed. Our experimental results show that this semantic burst detection in adaptive time windows is efficient and effectiveness for processing in both real time data stream and offline data stream.