Textual analysis of belgian newspapers with topic models

Details

Supervisors

Schaus, Pierre

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] en science des données, orientation technologies de l'information, à finalité spécialisée

Abstract

The objective of this master's thesis is to perform a textual analysis of various Belgian newspapers in order to compare and contrast their content. This analysis is facilitated by Latent Dirichlet Allocation, an unsupervised topic model capable of learning topics from text. Initially, we present the datasets we created for this work and the preprocessing steps employed to train the models effectively. Then, we present Latent Dirichlet Allocation along with techniques for visualizing, evaluating, labeling and exploring such models. Finally, we utilize these methods to conduct a content comparison of the newspapers.