Abstract: Twitter is a social networking and microblogging service on which users interact with messages. It is one of the best examples of microblogging and has a 280-character limit for a tweet. Registered users on twitter can post, like, and retweet tweets, but unregistered users can only read them. It is used not only to communicate with friends but also to share real-world events. Event detection is a major research area in text mining. Social media data (specifically, twitter data) is easily available. Twitter is a major source of information about real-world events. In twitter hashtags and word limit ensures the concise representation of real-world events. In this work, a segmentation based model is used to detect real-world events. Hashtags are the most important segment in the event detection process. The method of event detection is to split each tweet and hashtags into segments. From these segments extract the bursty segments. Then bursty segments are clustered based on the similarity measures. Finally, these clusters are summarized to produce final event. The key features of the event detection system are hashtags, retweet count, user popularity, and follower count. Here hashtags are more important and giving more weight to improve the performance of the model. The event detection system uses a Wikipedia title file for indexing the segments. Events2012 dataset is used for event detection. The results show the events are real-world in most of the cases.
Keywords: Microblogging, Hashtags, Natural language processing, Segmentation.
|
DOI:
10.17148/IJARCCE.2020.9656
