Extracting Features from Social Media Networks Using Semantics

This paper focuses on the analysis of social media content generated by social networks (e.g. Twitter) in order to extract semantic features. By using text categorization to sort text feeds into categories of similar feeds, it has been proved to reduce the overhead that is required to retrieve these feeds and at the same time, it provides smaller pools in which further investigations can be made easier. The aim of this survey is to draw a user profile, by analysing his or her tweets. In this early stage of research, being a pre-processing phase, a dictionary based approach is considered. Moreover, the paper describes an algorithm used in analysing the text and its preliminary results. This paper is focusing to support research in Social Media exploration. Thus, it describes a tool useful for communication experts to analyse public speeches. So far, this tool gave promising results in inferring socio-political trends from social media content of public speakers. We also evaluated our experiment on Support Vector Machine (SVM) with 10-fold cross-validations.