The Research On Text Linguistic Steganographic Improvements Based On Natural Language

With the drastic development of computer application technology and Internet,it is possible to transfer massive information quickly and conveniently via the Internet.In this social background,the communication among people becomes much more varied and frequent.Simultaneously,the circumstances on how to guarantee the security of information transmission are increasingly rigorous.In recent years,it raises growing issues,such as network software works copyright disputes and illegal text theft.Specifically,for example,it is pretty difficult for people to find proper way to transmit their confidential information safely in the network.What's more,the cipher text generated by the encryption algorithm is also stalked and attacked by the third party as a reason of char disorder.Thus,how to improve the security and concealment of confidential information transmission is gradually being an urgent issue.The paper focuses on Text Linguistic Stenographic(TLS),which is aimed at providing a secure method to transmit confidential information.Furthermore,this paper tries to improve Text Linguistic Stenographic(TLS)based on natural language.The main tasks in this paper are as follows:(1)In order to solve the problem of synonym words' category and polysemy in the Text Linguistic Stenographic(TLS)based on synonymy substitution rules,this paper proposes a stenographic algorithm based on the vector distance of two-gram synonyms dependent synonyms(SAVS).Specifically,first of all,target-terms which have the same semantic meaning are chosen from WordNet sets as candidate synonyms.Secondly,with the use of the dependency parser,this algorithm analyzes the binary dependent words collocation relations among synonyms in the statement.Thirdly,this algorithm calculates the frequency of the binary dependent words collocation relations in accord with candidated synonyms in the corpusFourthly,this algorithm calculates the collocation importance between the degree of synonyms scale and two-gram dependent synonyms.Finally,this algorithm calculates the two-gram dependent vector distance between target-terms and candidate synonyms and obtain the synonymy substitution fitness.The results show that the algorithm not only improves the accuracy of synonym substitution in the context,but also can increase efficiency in resisting attacks made by detection algorithms using statistics features based on the synonyms pairing and relative frequency.(2)In order to keep the statistical characteristics of texts,a multiple choice question based on dictionary sorting(MCQS)is proposed.A series of MCQS from multiple choices questions are automatically selected by secret information to generate a stego text,and their options are then reordered to embed more information.At the same time,this paper mixes the SAVS with the MACQS to embed more secret information and then generate multiple choice text.Experimental results show that the text generated from mixed algorithm has a better attack-defending ability and robustness than existed SAVS.It has an advantage on a high-embedded ratio and resisting statistical analysis attack.The research work of this paper improves Text Linguistic Stenographic by solving the shortage of english synonymy substitution rules based on natural language information.Besides,it contributes to the further research on MCQS in practice and in theory.