Crawling and Analysis of Online Discussions in Major Slovak National Newspapers

In this paper we compare the structure of online discussions in two major Slovak national newspapers (Sme.sk and Cas.sk) for two categories, domestic news and international news. We also perform clustering analysis for one of them (Cas.sk). We visualize online discussions through participant profiles, which combine content-oriented features, such as the number of expletives or URL links in the participants' posts, with communication-oriented features, such as how many distinct people ever responded to them. Using five or six most effective features, we identify five relatively stable clusters on the analyzed domain, four of which generalize across the two categories. Each cluster represents a group of discussion participants with similar characteristics and could be interpreted as a kind of participant type or role. We deliberately defined our features in a language-independent way so that they are directly usable in other languages besides Slovak.