Linguistic Profiling for Authorship Recognition and Verification

Linguistic Profiling for Author Recognition and Verification Hans van Halteren Language and Speech, Univ. of Nijmegen P.O. Box 9103 NL-6500 HD, Nijmegen, The Netherlands hvh@let.kun.nl Abstract A new technique is introduced, linguistic profiling, in which large numbers of counts of linguistic features are used as a text profile, which can then be compared to average profiles for groups of texts. The technique proves to be quite effective for authorship verification and recogni- tion. The best parameter settings yield a False Accept Rate of 8.1% at a False Re- ject Rate equal to zero for the verification task on a test corpus of student essays, and a 99.4% 2-way recognition accuracy on the same corpus. 1 Introduction There are several situations in language research or language engineering where we are in need of a specific type of extra-linguistic information about a text (document) and we would like to determine this information on the basis of lin- guistic properties of the text. Examples are the determination of the language variety or genre of a text, or a classification for document routing or information retrieval. For each of these applica- tions, techniques have been developed focusing on specific aspects of the text, often based on frequency counts of functions words in linguis- tics and of content words in language engineer- ing. In the technique we are introducing in this paper, linguistic profiling, we make no a priori choice for a specific type of word (or more complex fea- ture) to be counted. Instead, all possible features are included and it is determined by the statistics for the texts under consideration, and the distinc- tion to be made, how much weight, if any, each feature is to receive. Furthermore, the frequency counts are not used as absolute values, but rather as deviations from a norm, which is again deter- mined by the situation at hand. Our hypothesis is that this technique can bring a useful contribution to all tasks where it is necessary to distinguish one group of texts from another. In this paper the technique is tested for one specific type of group, namely the group of texts written by the same author. 2 Tasks and Application Scenarios Traditionally, work on the attribution of a text to an author is done in one of two environments. The first is that of literary and/or historical re- search where attribution is sought for a work of unknown origin (e.g. Mosteller & Wallace, 1984; Holmes, 1998). As secondary information gener- ally identifies potential authors, the task is au- thorship recognition: selection of one author from a set of known authors. Then there is forensic linguistics, where it needs to be determined if a suspect did or did not write a specific, probably incriminating, text (e.g. Broeders, 2001; Chaski, 2001). Here the task is authorship verification: confirming or denying authorship by a single known author. We would like to focus on a third environment, viz. that of the handling of large numbers of student essays. For some …

Content truncated to 500 words