Authorship Attribution of English Poetry using Sentiment Analysis

Abstract

We present a basic methodology and share some interesting results from experiments on using sentiment analysis for authorship attribution of poetry. We demonstrate that sentiment analysis can be effectively used to determine the authorship of poetic works given a sufficiently large training corpus. We also share some promising preliminary results from sentiment-analysis-based attribution of non-poetry works. Most results compare well with traditional authorship attribution approaches. Moreover, adding sentiment analysis to a traditional-feature-based ensemble classifier improved the accuracy of attribution. The strengths and limitations of our methodology and directions for further research are outlined at the end of the paper.