Author Profiles Prediction Using Syntactic and Content-Based Features

Abstract

In digital forensics, the forensic analysts raised the major questions about the details of the author of a document like identity, demographic information of authors and the documents which were related these documents. To answer these questions, the researchers proposed a new research field of stylometry which uses the set of linguistic features and machine learning algorithms. Information extraction from the textual documents has become a popular research area in the last few years to know the details of the authors. In this context, author profiling is one research area concentrated by the several researchers to know the authors’ demographic profiles like age, gender, and location by examining their style of writing. Several researchers proposed various types of stylistic features to analyze the style of the authors writing. In this paper, the experiment was performed with combination of syntactic features and content-based features. Various machine learning classifiers were used to evaluate the performance of the prediction of gender of reviews dataset. The proposed method achieved best accuracy for profiles prediction in author profiling.