Finding summaries to obtain event phrases from streaming Microblogs using Word Co-occurrence Network

Event detection from social media data has gained much momentum in research community. Existing event detection techniques are not scalable over less frequently discussed events. To handle this, BArank was proposed to identify keyphrases from topic specific Twitter data. Another concern for event detection from Microblogs is handling streaming data. Real-time Tweets are handled using Hadoop platform, or by referring old and new corpus which are complex and computationally expensive. However, handling streaming data depends upon the event detection technique. In this research work, different parameters are proposed to obtain discrete set of Tweets. The resulting phrases from BArank are ranked using Multi-Criteria Decision Making approach of Analytical Hierarchical Process (AHP) to identify events. Experimental results have proved high effectiveness of the proposed approach for FA Cup dataset with improved keyword-precision and keyword-recall without affecting topic-recall