Better Features Sets for Authorship Attribution of Short Messages

Abstract

Authorship authentication analysis can help to display information about the writers of messages by analyzing the writing styles. Previous researches in the authorship authentication were showed that generally people have their unique stylistic discriminators and characteristics, just like their fingerprints or signature. In this concept, researchers are developing different analysis features and techniques and have gained remarkable results in the authorship identification research field.
Authorship authentication of online messages became an outstanding research topic in the last decades because of internet usage growth. One of the problems of authorship authentication analysis regarding online sources is short messages usage. Author identification techniques are started to be applied to short and informal texts in last decade and get very significant results.
Authorship authentication is one of the security concerns in social network and in this research we will study how to authenticate a user by the writing style in a short text posted on Twitter. The effects of different feature sets and sample sizes are evaluated in the research.