A Survey on Stylometric Text Features

Abstract

Ways of individual style expression in a natural language include amongst other things stylometric features. These can be automatically detected with the use of computational linguistics methods. In this survey we systematize the recent studies devoted to extraction and application of stylometric features in solving natural language processing tasks: authorship attribution, authorship verification, style change detection, au- thorship profiling, and text classification by genre and sentiment. For that purpose we define stylometric feature categories that provide for the most effective solutions, discuss reasons for their successful application, touch upon the limitations of approaches based on their application, and make suggestions for future research.

Keywords

• natural language processing
• stylometric features
• stylometry
• authorship attribution
• authorship verification
• style change detection
• authorship profiling
• text classification by genre
• sentiment classification
• survey