Multi-author document decomposition based on authorship

Publication Type:
Issue Date:
Full metadata record
Decomposing a document written by more than one author into sentences based on authorship is of great significance due to the increasing demand for plagiarism detection, forensic analysis, civil law (i.e., disputed copyright issues) and intelligence issues that involves disputed anonymous documents. Among the existing studies for document decomposition, some were limited by specific languages, according to topics or restricted to a document of two authors, and their accuracies have big rooms for improvement. In this thesis, we propose novel approaches for decomposition of a multi-author document written in any language disregarding to topics, based on a Naive-Bayesian model and Hidden Markov Model (HMM). The proposed approaches of the Naive-Bayesian model aim to exploit the difference in its posterior probability to improve the performance of decomposition. Two main procedures are proposed based on Naive-Bayesian model, and they are Segment Elicitation procedure and Probability Indication Procedure. The segment elicitation procedure is proposed to form a strong labeled training dataset. The probability indication procedure is developed to improve the purity of the sentence decomposition. The proposed approaches of the HMM strive to exploit the contextual correlation hidden among sentences when determining their authorships. In this thesis, it is for the first time the sequential patterns hidden among document elements is considered for such a problem. To build and learn the HMM, a new unsupervised learning method is proposed to estimate its initial parameters. The proposed frameworks do not require the availability of any information of authors or document's context other than how many authors have contributed to writing the document. The effectiveness of the proposed algorithms is proved using benchmark datasets which are widely used for authorship analysis of documents. Furthermore, scientific papers are used to demonstrate the performance of the proposed approaches on authentic documents. Comparisons with recent state-the-art approaches are also presented to demonstrate the significance of our new ideas and the superior performance of the proposed approaches.
Please use this identifier to cite or link to this item: