Real-time progressive event summarization and sentiment analysis on evolutionary tweet stream

In the recent years, the popularity of Social Networking Sites (SNS) has increased immensely. So the numbers of tweets are generated at a huge rate immediately after the occurrence of the event. The users of the SNS always willing to get a compact understanding of opinions of people about a specific event without reading all tweets. This paper mainly focuses on mining of trending topics on twitter, analyzing the sentiments and obtaining a summary on the topic. The system attempts to create groups of tweets on a particular topic based on content similarity and then provides concise opinion summary for each group. We also determine sentiments of people related to the event. Sentiments are classified into three classes positive, negative and neutral using “Sentiwordnet dictionary”. As the different user can request the summary at any instant, the traditional event summarization and clustering techniques cannot fulfill the real-time necessity of this application. So, the summarization issue for the twitter's data is designed as progressive clustering and sentiment analysis task. Finally, the visualization interface gives a brief overview of various opinions about the event.