Rivindu Perera


A Tweet Classification Model Based on Dynamic and Static Component Topic Vectors

Abstract
This paper presents an unsupervised architecture for retrieving and ranking conceptually related tweets which can be used in real time. We present a model for ranking tweets with respect to topic relevance in order to improve the accuracy of information extraction. The proposed architecture uses concept enrichment from a knowledge source in order to expand the concept beyond the search keywords. The enriched concept is used to determine similarity levels between tweets and the given concept followed by a ranking of those tweets based on different similarity values. Tweets above a certain similarity threshold are considered as useful for providing relevant information (this is not part of this paper). We obtained precision values up to 0.81 and F values up to 0.61 for a tweet corpus of 2400 Tweets on the topic related to 2014 NZ general elections.