A computational analysis of the online Cypriot political discourse – interpreting big data

Abstract

Due to the development of the Internet, the online information sharing is done with a high speed. Different tasks can be performed through it, such as information retrieving and socialising by using different social media. The amount of online information is enormous and constantly increasing in a rapid manner. Of high importance is information published that is related to politics. Politicians use different social media platforms such as Twitter to express their views and opinions on different issues. Moreover, most of the news media outlets use websites to communicate their articles to the masses. The aim of this study is to create a system that follows and analyses the news in the political sphere in Cyprus in real time through the utilization of big-data techniques and approaches. To be more precise, the system will collect different politic related documents uploaded in Twitter and/or in online newspapers and analyse them in real time. The analysis of the data will be achieved using machine learning techniques. Specifically, the main technique used is the so called Latent Dirichlet Allocation (LDA). It is an unsupervised technique for clustering big amount of information to create groups of similar documents. The system proposed to be developed will be online and accessible through the internet. The main features of the system will be to allow its users to monitor the Cyprus political sphere by creating different analyses in different time windows. Finally, the users will be able to observe the sentiment for each topic in each analysis which indicates the general vibe around the topic.