语义空间下基于情感表达的生成式文本隐写方法

doi:10.11959/j.issn.1000-436x.2023045

通信学报 ›› 2023, Vol. 44 ›› Issue (4): 176-186.doi: 10.11959/j.issn.1000-436x.2023045

语义空间下基于情感表达的生成式文本隐写方法

刘玉玲¹, 王翠林¹, 付章杰²

¹ 湖南大学信息科学与工程学院，湖南长沙 410082
² 南京信息工程大学计算机学院、软件学院、网络空间安全学院，江苏南京 210044

修回日期:2023-01-04 出版日期:2023-04-25 发布日期:2023-04-01
作者简介:刘玉玲（1980- ），女，湖南宁乡人，博士，湖南大学副教授、博士生导师，主要研究方向为多媒体内容安全、保密技术、自然语言处理等
王翠林（1999- ），女，土家族，湖南湘西人，湖南大学硕士生，主要研究方向为文本内容安全、自然语言处理等
付章杰（1983- ），男，河南南阳人，博士，南京信息工程大学教授、博士生导师，主要研究方向为人工智能安全、区块链安全、数字取证等
基金资助:
国家自然科学基金资助项目(61872134);教育部科技发展中心基金资助项目(2019J01020);长沙市科技计划基金资助项目(Kh2004004);湖南省交通运输厅科技计划基金资助项目(201935)

Generative text steganography method based on emotional expression in semantic space

Yuling LIU¹, Cuilin WANG¹, Zhangjie FU²

¹ School of Computer Science and Electronic Engineering, Hunan University, Changsha 410082, China
² School of Computer Science, Nanjing University of Information Science and Technology, Nanjing 210044, China

Revised:2023-01-04 Online:2023-04-25 Published:2023-04-01
Supported by:
The National Natural Science Foundation of China(61872134);The Science and Technology Development Center of the Ministry of Education(2019J01020);Science and Technology Project of Changsha City(Kh2004004);Science and Technology Project of Transport Department of Hunan Province(201935)

摘要/Abstract

摘要：

针对现有生成式文本隐写方法存在的“过度优化”文本质量以及生成的隐写文本在语义表达上缺乏约束等问题，提出了一种在语义空间下基于情感表达的生成式文本隐写方法。该方法利用新媒体平台提供的情景融合的伪装场景，研究如何利用无监督抽取模型从原始数据集中抽取情感表达组合候选集合，并基于改进的二部图排序算法对情感表达组合候选集合进行排序，得到情感表达组合集合；然后将其映射到语义空间，实现基于情感表达组合生成用户观点的同时嵌入秘密信息。实验结果表明，与同类语义空间下生成式文本隐写方法相比，所提方法生成的含密商品评论的困惑度最低可达10.536，且含密商品评论与主题具有较强相关性，进一步保证了隐写文本的认知隐蔽性，同时所提方法还可有效地用于安全保密通信领域，能够避免发送方被追踪溯源和关联分析。

关键词: 生成式文本隐写, 语义空间, 无监督抽取模型, 情感表达

Abstract:

Aiming at the problems that “over optimizing” the quality of steganographic text and lack of constraints on the semantic expression of the generated steganographic text in existing generative text steganography methods, a generative text steganography method was proposed based on emotional expression in semantic space.In order to make use of the scene fusion provided by the new media platform to obtain many camouflage scenes, the focus was how to use the unsupervised extraction model to extract the emotional expression combination candidate set from the original data set, then sort the candidate set of emotional expression combinations based on the improved bipartite graph sorting algorithm to obtain the emotional expression combination set, map them to the semantic space, and then implement embedding secret information while generating the user’s opinions based on the emotion expression combinations.Experimental results show that, compared with the existing generative text steganography methods in semantic space, the product reviews generated by the proposed method have a minimum perplexity of 10.536, and have a strong correlation with the chosen product, which can further guarantee the cognitive concealment of steganographic texts.At the same time, the proposed method can also be effectively used in the field of secure and confidential communication, and can avoid the senders being traced and analyzed.

Key words: generative text steganography, semantic space, unsupervised extraction model, emotional expression

中图分类号:

TP309

刘玉玲, 王翠林, 付章杰. 语义空间下基于情感表达的生成式文本隐写方法[J]. 通信学报, 2023, 44(4): 176-186.

Yuling LIU, Cuilin WANG, Zhangjie FU. Generative text steganography method based on emotional expression in semantic space[J]. Journal on Communications, 2023, 44(4): 176-186.

图/表 15

图1

图2

表1

图3

表2

表3

图4

表4

表5

表6

图5

表7

图6

图7

图8

参考文献 26

[1]	PETITCOLAS F A P , ANDERSON R J , KUHN M G . Information hiding-a survey[J]. Proceedings of the IEEE, 1999,87(7): 1062-1078.
[2]	吴国华, 龚礼春, 袁理锋 ,等. 中文文本信息隐藏研究进展[J]. 通信学报, 2019,40(9): 145-156.
	WU G H , GONG L C , YUAN L F ,et al. Review of information hiding on Chinese text[J]. Journal on Communications, 2019,40(9): 145-156.
[3]	张祯, 倪嘉铭, 姚晔 ,等. 基于同义词扩展和标签传递机制的文本无载体信息隐藏方法[J]. 通信学报, 2021,42(9): 173-183.
	ZHANG Z , NI J M , YAO Y ,et al. Text coverless information hiding method based on synonyms expansion and label delivery mechanism[J]. Journal on Communications, 2021,42(9): 173-183.
[4]	WANG C L , LIU Y L , TONG Y J ,et al. GAN-GLS:generative lyric steganography based on generative adversarial networks[J]. Computers,Materials ＆ Continua, 2021,69(1): 1375-1390.
[5]	WAYNER P . Mimic functions[J]. Cryptologia, 1992,16(3): 193-214.
[6]	CHAPMAN M , DAVIDA G I , RENNHARD M . A practical and effective approach to large-scale automated linguistic steganography[C]// Proceedings of the 4th International Conference on Information Security. New York:ACM Press, 2001: 156-165.
[7]	DESOKY A . Nostega:a novel noiseless steganography paradigm[J]. Journal of Digital Forensic Practice, 2008,2(3): 132-139.
[8]	LUO Y B , HUANG Y F , LI F F ,et al. Text steganography based on ci-poetry generation using Markov chain model[J]. KSII Transactions on Internet and Information Systems, 2016,10(9): 4568-4584.
[9]	YI X Y , LI R Y , SUN M S . Generating Chinese classical poems with RNN encoder-decoder[C]// International Symposium on Natural Language Processing Based on Naturally Annotated Big Data,China National Conference on Chinese Computational Linguistics. Berlin:Springer, 2017: 211-223.
[10]	YANG Z L , GUO X Q , CHEN Z M ,et al. RNN-Stega:linguistic steganography based on recurrent neural networks[J]. IEEE Transactions on Information Forensics and Security, 2019,14(5): 1280-1295.
[11]	ZIEGLER Z , DENG Y T , RUSH A . Neural linguistic steganography[C]// Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). Stroudsburg:Association for Computational Linguistics, 2019: 1210-1215.
[12]	XIANG L Y , YANG S H , LIU Y H ,et al. Novel linguistic steganography based on character-level text generation[J]. Mathematics, 2020,8(9): 1558.
[13]	NAKAJIMA T V , KER A D . The syndrome-trellis sampler for generative steganography[C]// Proceedings of 2020 IEEE International Workshop on Information Forensics and Security (WIFS). Piscataway:IEEE Press, 2021: 1-6.
[14]	ZHOU X J , PENG W L , YANG B Y ,et al. Linguistic steganography based on adaptive probability distribution[J]. IEEE Transactions on Dependable and Secure Computing, 2022,19(5): 2982-2997.
[15]	石凤贵 . 基于 jieba 中文分词的中文文本语料预处理模块实现[J]. 电脑知识与技术, 2020,16(14): 248-251,257.
	SHI F G . Realization of Chinese text corpus preprocessing module based on jieba Chinese word segmentation[J]. Computer Knowledge and Technology, 2020,16(14): 248-251,257.
[16]	卢奇, 陈文亮 . 大规模中文实体情感知识的自动获取[J]. 中文信息学报, 2018,32(8): 32-41.
	LU Q , CHEN W L . Automatically building a large scale dictionary of Chinese entity sentiment expressions[J]. Journal of Chinese Information Processing, 2018,32(8): 32-41.
[17]	MIKOLOV T , CHEN K , CORRADO G ,et al. Efficient estimation of word representations in vector space[J]. arXiv Preprint,arXiv:1301.3781, 2013.
[18]	YANG Z L , GONG B T , LI Y M ,et al. Graph-Stega:semantic controllable steganographic text generation guided by knowledge graph[J]. arXiv Preprint,arXiv:2006.08339, 2020.
[19]	ZHANG S Y , YANG Z L , YANG J S ,et al. Provably secure generative linguistic steganography[C]// Proceedings of Findings of the Association for Computational Linguistics:ACL-IJCNLP 2021. Stroudsburg:Association for Computational Linguistics, 2021: 3046-3055.
[20]	YANG Z L , ZHANG S Y , HU Y T ,et al. VAE-Stega:linguistic steganography based on variational auto-encoder[J]. IEEE Transactions on Information Forensics and Security, 2021,16: 880-895.
[21]	JELINEK F , MERCER R L , BAHL L R ,et al. Perplexity—a measure of the difficulty of speech recognition tasks[J]. The Journal of the Acoustical Society of America, 1977,62(S1): S63.
[22]	FANG T N , JAGGI M , ARGYRAKI K . Generating steganographic text with LSTMs[C]// Proceedings of ACL 2017,Student Research Workshop. Stroudsburg:Association for Computational Linguistics, 2017: 100-106.
[23]	MAAS A L , DALY R E , PHAM P T ,et al. Learning word vectors for sentiment analysis[C]// Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics:Human Language Technologies. New York:ACM Press, 2011: 142-150.
[24]	ZHANG S Y , YANG Z L , YANG J S ,et al. Linguistic steganography:from symbolic space to semantic space[J]. IEEE Signal Processing Letters, 2021,28: 11-15.
[25]	ZHOU H Y , WANG F , TAO P . T-distributed stochastic neighbor embedding method with the least information loss for macromolecular simulations[J]. Journal of Chemical Theory and Computation, 2018,14(11): 5499-5510.
[26]	PAPINENI K , ROUKOS S , WARD T ,et al. BLEU:a method for automatic evaluation of machine translation[C]// Proceedings of the 40th Annual Meeting on Association for Computational Linguistics. New York:ACM Press, 2002: 311-318.

数据集	对象属性列表/个	连接词/个
网易考拉	73 321	84 923
淘宝	13 215	20 329
京东	24 312	34 578

符号	定义
K	通信双方选定的商品
O	商品列表集合
Size	商品列表的长度
P	选定的商品属性
Q	选定的情感组合
Extraction_model	无监督提取函数
dict	数据字典
Secret_information	待嵌入的秘密信息
GetIndex	获取秘密信息索引下标函数
I	秘密信息的索引下标集合
L	索引下标集合的长度
C	商品评论列表集合
C＇	含密商品评论

左邻字集合	右邻字集合
个	的
的	盒
前	的
的	盒

数据集	商品评论句子数/句	字符数/B	平均每句字符数/B	商品类别/种
网易考拉	55 456	2 661 954	48.01	25
淘宝	14 232	687 690	48.32	25
京东	23 911	1 182 160	49.44	25

方法	嵌入容量/(bit·word^-1)	PPL
	1.000	30.665
LSTM	2.000	10.027
	3.000	74.543
	1.000	45.115
VAE-Stega (LSTM-LSTM) (HC)	1.863	49.511
	2.577	59.532
	1.000	30.266
VAE-Stega (BERT-LSTM) (AC)	1.866	36.349
	2.596	40.832
	1.000	30.266
VAE-Stega (BERT-LSTM)(HC)	1.866	36.349
	2.596	40.832
	1.000	20.915
RNN-Stega (HC)	1.845	24.839
	2.565	29.187
	1.000	20.341
本文方法	2.000	20.933
	3.000	20.431

语义空间下基于情感表达的生成式文本隐写方法

Generative text steganography method based on emotional expression in semantic space

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 15

参考文献 26

相关文章 1

Metrics

推荐阅读 0

t	本文方法	CTRL
4	10.536～10.900	16.003～16.541
8	11.453～11.543	18.661～18.983
10	13.334～13.454	18.901～19.345
16	14.342～14.432	19.494～19.834

t	1-gram	2-gram	3-gram	4-gram
4	0.834	0.764	0.642	0.554
10	0.778	0.704	0.597	0.441