A Mathematical Formulation of Classified Variables for Enhanced Detection in Natural Language Steganalysis

Abstract

Steganalysis, which is the counterpart to steganography, is an art and science that is devoted to detecting hidden information that has been concealed within seemingly harmless digital media. With the advancement in technology nowadays, the techniques used in steganography also become more complex, making it necessary to continuously improve steganalysis methods to keep up with emerging threats and ensure digital data's confidentiality, authenticity, and privacy. Steganalysts face the challenge of uncovering hidden information within covert media, requiring analysis of both original and altered media. While efficient steganalysis tools exist for specific approaches, devising a universal solution for all steganography techniques remains challenging. With the utilization of the formulated key, this paper aims to classify and analyze variables used in the steganalytic system. Thus, three views of classified variables are presented to address the pattern of detection. These are known as trade-off value-based, probability of character variable, and Support Vector Machine-based (SVM-based). Hence, it is expected that this scheme will become one of the alternative ways to enhance the steganalytic system for discovering the hidden message in communication. Furthermore, it is suggested the necessity for enhanced steganalysis techniques and tools, as well as the significance of academic and professional education in this area.