Authorship attribution and feature testing for short Chinese emails

作者：Zhang, Shaomin^*

来源：International Journal of Speech Language and the Law, 2016, 23(1): 71-97.

DOI：10.1558/ijsll.v23i1.20300

• AUTHORSHIP ATTRIBUTION
• AUTHORSHIP FEATURES
• PRAGMATIC FEATURES
• DISCOURSE SEMANTIC FEATURES
• DISCOURSE INFORMATION FEATURES
• SHORT CHINESE EMAILS

Features used in English-based authorship attribution demonstrate some constraints when used in authorship attribution of Chinese texts, even though they shed much light on it. Therefore, authorship attribution for short Chinese texts free of handwritten documents will help to promote the progress of legislation in China. This study aims to explore and test some features in pragmatic, discourse semantic and discourse information features for authorship attribution of short Chinese emails, which are representative of communication tools. It is hoped that some effective features can be found for the attribution of short Chinese emails. The texts used in the study include 72 short emails written by six authors. All the possible 57 combinations of the six authors are tested and attributed based on the extracted features. Discriminant analysis is employed, and the results demonstrate significant predictions in all the tests. It is concluded that the extracted features in pragmatics, discourse semantics and discourse information can significantly distinguish short Chinese emails, and the suggested number of suspect authors for authorship attribution of short emails should not exceed five.

• 出版日期2016
• 单位广东技术师范大学