Research on information security in text emotional steganography based on machine learning

• Published in: Enterprise information systems Vol. 15; no. 7; pp. 984 - 1001
• Main Authors: Li, Fufang, Tang, Han, Zou, Yukang, Huang, Yongfeng, Feng, Yuanyong, Peng, Lingxi
• Format: Journal Article
• Language: English
• Published: Taylor & Francis 09.08.2021
• Subjects: Distributed Hash Table (DHT); emotional information steganography (hiding) and analysis; emotional machine learning; matrix encoding; Network on Chip (NoC) based intelligent autonomous application; Sensitive information security
• ISSN: 1751-7575; 1751-7583
• DOI: 10.1080/17517575.2020.1720827

To effectively protect sensitive information, this paper proposes a method of text Emotional modulation steganography based on machine learning. Firstly, we explore an intelligent dynamic expansion method for text emotional lexicon based on deep learning in detail. Secondly, the cosine similarity algorithm is used to combine the two most similar emotional words into emotional word pairs, which is the basis of the proposed approach. Thirdly, we combine emotional word pairs with matrix encoding algorithm to calculate the minimum modification unit, so as to minimize the rewriting of carrier text, and hence improve the privacy, embedding rate and capacity of sensitive information. Fourthly, we evaluate and validate the proposed algorithms through several experiments. Finally, the design of a NoC based organizational covert and secure autonomous intelligent online team work system is carried out. Experiments show that the efficiency, security, concealment and robustness of the proposed algorithms are sound.