We explore techniques to maximize the effectiveness of discourse information in the task of authorship attribution. We present a novel method to embed discourse features in a Convolutional Neural Network text classifier, which achieves a state-of-the-art result by a substantial margin. We empirically investigate several featurization methods to understand the conditions under which discourse features contribute non-trivial performance gains, and analyze discourse embeddings.
@article{arxiv.1709.02271,
title = {Leveraging Discourse Information Effectively for Authorship Attribution},
author = {Su Wang and Elisa Ferracane and Raymond J. Mooney},
journal= {arXiv preprint arXiv:1709.02271},
year = {2017}
}