Authorship Analysis with Machine Learning

Abstract

The Internet has evolved with the development of textual data. Crimes related to the use of textual data on the Internet have also increased. Attribution is a technique of text classification used to identify an unknown file by analyzing files from multiple authors. Authorship analysis is widely studied for identifying individuals based on their writing style extracted from their written discourses. Most studies reveal that humans leave (consciously or unconsciously) their personality traits and sociolinguistic characteristics in their writing. Authorship analysis methods range from manual, statistical, computational, and machine learning to deep learning. In the current Industry Evolution 4.0 (4IR) era, the digital world contains large amounts of data, including social media, mobile, network, and Internet of Things (IoT). The key to intelligence is analyzing this data and developing appropriate automated and smart applications using artificial intelligence (AI) and machine learning (ML) algorithms. According to the perspective of data computing and analysis, AI and ML have developed rapidly, and this data generally allows applications to operate intelligently. Moreover, ML typically provides a system that can automatically learn from experience without any special programming. Hence, it is often called the latest technology of the 4IR. This entryr studies machine learning, including supervised, unsupervised, semi-supervised, and reinforcement learning for authorship analysis. Recently developed learning models such as recurrent neural networks and bidirectional encoder representations from transformers are also explored. Further, this entry presents the open issues and emerging challenges associated with authorship learning models.