TweetSemMiner: A Meta-Topic Identification Model for Twitter Using Semantic Analysis

Abstract

The information contained in Social Networks has become increasingly important over the last few years. Inside this field, Twitter is one of the main current information sources, produced by the comments and contents that their users interchange. This information is usually noisy, however, there are some hidden patterns that can be extracted such as trends, opinions, sentiments, etc. These patterns are useful to generate users communities, which can be focused, for example, on marketing campaigns. Nevertheless, the identification process is usually blind, difficulting this information extaction. Based on this idea, this work pretends to extract relevant data from Twitter. In order to achieve this goal, we have desgined a system, called TweetSemMiner, to classify user comments (or tweets) using general topics (or meta-topics). There are several works devoted to analize social networks, however, only Topic Detection techniques have been applied in this context. This paper provides a new approach to the problem of classification using semantic analysis. The system has been developed focused on the detection of a single meta-topic and uses techniques such as Latent Semantic Analysis (LSA) combined with semantic queries in DBpedia, in order to obtain some results which can be used to analyze the effectiveness of the model. We have tested the model using real users, whose comments were subsequently evaluated to check the effectiveness of this approach.