Information Hiding And Analysis Based On Text Emotional Features

As an important branch of information security research,information hiding mainly focuses on hiding secret information in vectors such as video,image,audio and text to achieve the purpose of secret communication.Compared with other types of carriers,text has the advantages of intuitive processing and large amount of data,but at the same time it is difficult to hide more information in the text due to low text redundancy.At present,the most researches at home and abroad are semantic-based text information hiding.The semanticbased text information hiding method is a hidden method that does not change the semantics by transforming the words themselves.Among them,the method of synonym substitution has achieved more results.The traditional synonym substitution method relies on the synonym dictionary.The most commonly used synonym dictionary is Synonym Forest.However,with the change of the language environment,many words in Synonym Forest have not been able to adapt well to the current research environment.With the rapid development of Internet Users’ Generate Content(UGC),a large number of new online words have been invented,and the expansion of new words based on the Internet has gradually developed into a research hotspot.The main idea of information hiding based on synonym substitution is to find the synonym of the carrier text in the dictionary,and modify the words of the carrier text according to the bit information embedded in the need to achieve the purpose of embedding the secret information.However,too much embedded information will cause the carrier to be manipulated and encoded more frequently,which will result in lower embedding rate and easy detection by statistical steganographic analysis tools.Based on the above problems,this thesis proposes a method of dynamically expanding the sentiment dictionary.The specific rules are used to identify the sentiment words in the corpus.The word vector tool word2 vec is used to learn the semantic relationship of the words in the corpus.The cosine similarity algorithm will be the most similar.Two sentiment words are combined into a sentiment word pair.Finally,the sentiment word pair is combined with the matrix coding algorithm,and the minimum modification unit is calculated to reduce the rewriting of the carrier,and the embedding rate of the text hiding is increased.In this thesis,the text perplexity is used as the evaluation index of the steganography algorithm,and the detection result of the support vector machine for the steganographic text is taken as the algorithm performance index.The experimental analysis shows that the extended sentiment dictionary combined with matrix coding can improve the embedding efficiency of the steganography algorithm,reduce the possibility of being detected by statistical analysis tools,and improve the security of secret information.