Short text(tweet) is one of most predominant text messaging and dissemination services being used by millions to gen- erate billions of atmost 140 characters messages. Tweets can be processed to determine events, trends, and to pro- vide multiple services. One of the challenges is to determine what a particlular tweet covers. Because of short text many a times, tweets are incomplete and ambiguous. Therefore, extracting some semantics from tweets is a challenge. In this paper, we present a framework to extract likely se- mantics from a tweet by considering syntax and semantics aspects along with domain specific inputs. We present re- sults of our solution by providing likely semantics for a tweet, and show that our results are able to match true semantics of the tweet most of the time. We provide an insight into few challenging problems that need to be further addressed to improve our solution.
Association of concepts to tweets can have many applications:-
- Filter for identifying tweets of a particular category/entity/domain
- Microblog advertising
- Identification of trending topics
Click on below links for results on various datasets