Automatically Identifying Pseudepigraphic Texts

Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing, pages 1449–1454, Seattle, Washington, USA, 18-21 October 2013. c⃝2013 Association for Computational Linguistics Automatically Identifying Pseudepigraphic Texts Moshe Koppel Bar Ilan University Ramat-Gan, 52900, Israel moishk@gmail.com Shachar Seidman Bar Ilan University Ramat-Gan, 52900, Israel shachar9@gmail.com Abstract The identification of pseudepigraphic texts – texts not written by the authors to which they are attributed – has important historical, fo- rensic and commercial applications. We in- troduce an unsupervised technique for identi- fying pseudepigrapha. The idea is to identify textual outliers in a corpus based on the pair- wise similarities of all documents in the cor- pus. The crucial point is that document simi- larity not be measured in any of the standard ways but rather be based on the output of a re- cently introduced algorithm for authorship ve- rification. The proposed method strongly outperforms existing techniques in systematic experiments on a blog corpus. 1 Introduction The Shakespeare attribution problem is centuries old and shows no signs of abating. Some scholars argue that some, or even all, of Shakespeare’s works were not actually written by him. The most mainstream theory – and the one that interests us here – is that most of the works were written by Shakespeare, but that several of them were not. Could modern methods of computational author- ship attribution be used to detect which, if any, of the works attributed to Shakespeare were not writ- ten by him? More generally, this paper deals with the unsu- pervised problem of detecting pseudepigrapha: documents in a supposedly single-author corpus that were not actually written by the corpus’s pre- sumed author. Studies as early as Mendenhall (1887), have observed that texts by a single author tend to be somewhat homogeneous in style. If this is indeed the case, we would expect that pseudepi- grapha would be detectable as outliers. Identifying such outlier texts is, of course, a special case of general outlier identification, one of the central tasks of statistics. We will thus consider the pseudepigrapha problem in the context of the more general outlier detection problem. Typically, research on textual outliers assumes that we have a corpus of known authentic docu- ments and are asked to decide if a specified other document is authentic or not (Juola and Stamata- tos, 2013). One crucial aspect of our problem is that we do not assume that any specific text in a corpus is known a priori to be authentic or pseude- pigraphic; we can assume only that most of the documents in the corpus are authentic. The method we introduce in this paper builds on the approach of Koppel and Winter (2013) for de- termining if two documents are by the same au- thor. We apply that method to every pair of documents in a corpus and use properties of the resulting adjacency graph to identify outliers. In the following section, we briefly outline previous work. In Section 3 we provide a framework for outlier detection and in …

Content truncated to 500 words