Real World Event Identification Twitter

The messages posted on twitter by users are a powerful way of sharing information on the web. These short messages can be about a variety of events which are occurring in real time and thus it makes a good source for analysing events in real time. In this project, we are trying to analyse a stream of twitter messages and distinguish between the messages which are about real world events and about non events. We are streaming the tweets using the twitter streaming api and then preprocessing the tweets and then finding out the top words which occur in every batch of tweets and then running the LDA algorithm on it to find out the trending topics of the hour. When this algorithm is extended to a large scale of twitter messages, we see the potential of clustering twitter messages in real world event identification.