Automatic categorisation of comments in social news websites

The use of the social web has brought a series of changes in the way how content is created. In particular, social news sites link stories and the different users can comment them. In this paper, we propose a new method based on different features extracted from the text able to categorise the comments. To this end, we use a combination of statistical, syntactic and opinion features and machine-learning classifiers to classify a comment within three different categorisation types: the focus of the comment, the type of information contained in the comment and the controversy level of the comment. We validate our approach with data from ‘Menéame’, a popular Spanish social news site.

► We propose a new method for representing comments in social news websites.
► We describel this machine-learning-based method for categorising comments in social news sites.
► The method can achieve high accuracy rates in three different classification tasks.