Deep Combination of Stylometry Features for Authorship Analysis

Authorship Analysis (AA) is a process aim to extract information about an author from his/her writings. To analyze whether two anonymous short texts were written by the same author, we propose a combination of stylometry features from different categories in different progress. The majority of the previous AA studies use many stylometry features from different categories together at the beginning of a solution as a pre-processing step. During the learning process, no category-specific operations are performed; all categories used are evaluated equally. However, the proposed approach has a separate learning process for each feature category and combines these processes at the decision phase by using a Combination of Deep Neural Networks (C-DNN). To evaluate the Authorship Verification (AV) performance of the proposed approach, we designed and implemented a problem-specific Deep Neural Network (DNN) for each stylometry category we used. Experiments were conducted on two English public datasets. The results show that the proposed approach significantly improves the generalization ability and robustness of the solutions, and also have better accuracy than the single DNNs.