In this work we study the problem of Authorship Attribution for a novel set of documents, namely online chats. Although the problem of Authorship Attribution has been extensively investigated for different document types, from books to letters and from emails to blog posts, to the best of our knowledge this is the first study of Authorship Attribution for conversational documents (IRC chat logs) using statistical models. We experimentally demonstrate the unsuitability of the classical statistical models for conversational documents and propose a novel approach which is able to achieve a high accuracy rate (up to 95%) for hundreds of authors.
|Title of host publication||2013 International Conference on Social Computing (SocialCom)|
|Place of Publication||Piscataway, NJ|
|Publication status||Published - 2013|
|Event||International Conference on Social Computing, SocialCom 2013, SocialCom/PASSAT/BigData/EconCom/BioMedCom 2013, Washington, DC, USA, 8-14 September, 2013 - |
Duration: 1 Jan 2013 → …
|Conference||International Conference on Social Computing, SocialCom 2013, SocialCom/PASSAT/BigData/EconCom/BioMedCom 2013, Washington, DC, USA, 8-14 September, 2013|
|Period||1/01/13 → …|