Applications of computational methods of clustering and authorship attribution
MetadataShow full item record
By examining the textual features of a writing, we can gain insight into the stylometric choices of its author. This paper seeks to examine the problem of anonymous and pseudonymous texts by looking at computational methods of authorship attribution, the algorithms they use, and the linguistic features they examine. The strengths and weaknesses of the different algorithms, as they are applied to differing texts, are also discussed. Through this framework, we explore two very different problems in literature. First, the difference between poetic and lexical style, as well as their indication of authorship, is examined through the domain of Middle English morality plays, focusing on those plays of the Towneley Cycle thought to have been written and revised by the Wakefield Master. Second, the effect of source language and translation on the textual features of a story is examined through the domain of Vladimir Nabokov's short stories. This research focuses on the use of a lexomics-based clustering algorithm as well as the significance of its results to the question of authorship.
File:thesis.pdfMIME type:application/pdfFile Size:538.4Kb