Stance Detection Based on Ensembles of Classifiers

Abstract

A method of stance detection in text is proposed. This method is based on the machine learning of ensembles of classifiers. It is known that ensembles have advantages over individual classifiers, which often improves the quality of classification. An important issue is determining the classifiers that should be included in such an ensemble. The method of constructing ensembles proposed in this paper, which is based on a cross-validation procedure, makes it possible to optimize the parameters of the base classifiers, evaluate the effectiveness of each combination of classifiers included in the set, and select the optimal combination. For testing the proposed method, corpora of Russian language messages in Internet forums and the social network VKontakte have been formed. These messages concern three socially significant issues—vaccination of children, Unified State Exam, and human cloning. The experimental study shows the advantage of the proposed method over other classifiers.