Exploring syntactic and semantic features for authorship attribution

Haiyan Wu, Zhiqiang Zhang, Qingfeng Wu

Abstract: Highlights•N-Gram attention Network for feature extraction is proposed.•Multi-level linguistic features are applied to the task of Authorship attribution.•Combining those extracted features with Logistic Regression algorithm and comparing with baselines, experimental results show our proposed method is effective.•Finally, we analyzed each type of feature, especially the syntactic structure.