Authorship attribution for small texts: Literary and forensic experiments

The 11th International Rivers ymposium

生态环境科学

Ol’ga Feiguina ; Graeme Hirst

Others  :  http://CEUR-WS.org/Vol-276/paper3.pdf
PID  :  10496

学科分类：环境科学（综合）

来源: CEUR

【 摘 要 】

To capture syntactic structure as a feature for the classification of short texts by their authorship, we use the frequencies of bigrams in the stream of syntactic labels produced by a partial parser. We experimented on literary data (from the Brontë sisters) and simulated forensic data. Syntactic label bigrams were found to be helpful with the former but not the latter.