CN114648435A - Method, device and equipment for detecting watermark in text and storage medium - Google Patents
Method, device and equipment for detecting watermark in text and storage medium Download PDFInfo
- Publication number
- CN114648435A CN114648435A CN202011497653.7A CN202011497653A CN114648435A CN 114648435 A CN114648435 A CN 114648435A CN 202011497653 A CN202011497653 A CN 202011497653A CN 114648435 A CN114648435 A CN 114648435A
- Authority
- CN
- China
- Prior art keywords
- text
- watermark
- processed
- emotional
- ciphertext
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 73
- 230000002996 emotional effect Effects 0.000 claims abstract description 102
- 238000001514 detection method Methods 0.000 claims abstract description 33
- 238000006243 chemical reaction Methods 0.000 claims abstract description 27
- 230000008451 emotion Effects 0.000 claims description 95
- 239000013598 vector Substances 0.000 claims description 84
- 238000012545 processing Methods 0.000 claims description 31
- 230000015654 memory Effects 0.000 claims description 23
- 238000003079 width control Methods 0.000 claims description 20
- 238000010606 normalization Methods 0.000 claims description 18
- 238000000605 extraction Methods 0.000 claims description 10
- 230000009466 transformation Effects 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 5
- 238000005516 engineering process Methods 0.000 abstract description 15
- 230000007547 defect Effects 0.000 abstract 1
- 230000008569 process Effects 0.000 description 15
- 230000006403 short-term memory Effects 0.000 description 12
- 238000004364 calculation method Methods 0.000 description 10
- 230000007787 long-term memory Effects 0.000 description 10
- 230000005540 biological transmission Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 6
- 230000008859 change Effects 0.000 description 4
- 238000004891 communication Methods 0.000 description 4
- 230000007246 mechanism Effects 0.000 description 4
- 238000003672 processing method Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 238000012216 screening Methods 0.000 description 3
- 238000012952 Resampling Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000006835 compression Effects 0.000 description 2
- 238000007906 compression Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 230000007935 neutral effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 230000000149 penetrating effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/0021—Image watermarking
- G06T1/005—Robust watermarking, e.g. average attack or collusion attack resistant
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/10—Protecting distributed programs or content, e.g. vending or licensing of copyrighted material ; Digital rights management [DRM]
- G06F21/16—Program or content traceability, e.g. by watermarking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2201/00—General purpose image data processing
- G06T2201/005—Image watermarking
- G06T2201/0062—Embedding of the watermark in text images, e.g. watermarking text documents using letter skew, letter distance or row distance
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Technology Law (AREA)
- Computer Hardware Design (AREA)
- Computer Security & Cryptography (AREA)
- Image Processing (AREA)
- Editing Of Facsimile Originals (AREA)
Abstract
The application discloses a method, a device, equipment and a storage medium for detecting a watermark in a text, wherein the method comprises the following steps: extracting an emotional characteristic mask of a text to be processed; under the condition that the watermark identification exists in the text to be processed, acquiring a ciphertext watermark of the text to be processed, wherein the ciphertext watermark is formed by embedding the emotional characteristic mask into the text to be processed according to a conversion rule; determining the matching degree between the emotional characteristic mask and the ciphertext watermark; and carrying out ciphertext watermark detection on the text to be processed according to the matching degree. The embodiment provided by the application solves the problems that the ciphertext watermark detection technology of the digital text has various defects in the aspects of safety, transparency, robustness and the like and can not realize good balance on performance.
Description
Technical Field
The present application relates to the technical field of copyright protection, and relates to, but is not limited to, a method, an apparatus, a device, and a storage medium for detecting a watermark in a text.
Background
The digital watermarking technology in the related technology is a computer hiding technology based on a content and non-password mechanism, and some identification information is directly embedded into a digital carrier, so that the use value of the original carrier is not influenced while the information security is protected. Text watermarking is a research branch of digital watermarking, combines the technical characteristics of text watermarking, and solves the technical problems of text watermarking and existing technology in three related technologies as follows:
the watermark embedding and detecting technology based on the text format is characterized in that a text bitmap is converted into an image, and the watermark technology mature in the image field is directly applied to text watermarking. And meanwhile, the presentation form of the digital watermark is controlled by using the style characteristics of the text, such as segment spacing, character spacing and the like. The problems of the technology are as follows: the style characteristics cannot detect that the file is tampered, and the safety performance is low.
Secondly, watermark embedding and detecting technology based on text data can insert inconspicuous errors of spelling, syntax, punctuation and even content in the text to embed the watermark. The problems of the technology are as follows: the original file content can be changed, the semantic change of the text content can be caused, the transparency principle of the watermark is damaged, the stable text watermark can not be generated, when the data is slightly damaged, different text watermarks can be generated, the watermark detection result is influenced, and the robustness is not strong; the embedded watermark is not detectable, because the embedded watermark has the same text characteristic with the original carrier, an illegal interceptor cannot judge whether hidden information exists.
And thirdly, based on the watermark embedding and detecting technology of the text content, the text semantics can be analyzed according to the text content, the text content characteristics are calculated by combining a statistical method or a syntactic method, and a simulation function is written to generate secret information penetrating through the whole text to realize watermark embedding. The technology hides the watermark in the statistical rules or grammatical codes of the text. The problems of the technology are as follows: the content feature volume is large, and more redundant feature data can be brought to the original carrier.
In summary, the existing ciphertext watermark detection technology for digital text has disadvantages in security, transparency, robustness, etc., or cannot achieve good balance in performance.
Disclosure of Invention
In view of this, the present application provides a method, an apparatus, a device, and a storage medium for detecting a watermark in a text to solve at least one problem in the prior art, which at least solves the problem that the prior art has disadvantages in security, transparency, robustness, and the like, or cannot achieve a good balance in performance.
The technical scheme of the application is realized as follows:
in a first aspect, the present application provides a method for detecting a watermark in a text, including: extracting an emotional characteristic mask of a text to be processed; under the condition that the watermark identification exists in the text to be processed, acquiring a ciphertext watermark of the text to be processed, wherein the ciphertext watermark is formed by embedding the emotional characteristic mask into the text to be processed according to a conversion rule; determining the matching degree between the emotional characteristic mask and the ciphertext watermark; and carrying out ciphertext watermark detection on the text to be processed according to the matching degree.
In a second aspect, the present application provides an apparatus for detecting a watermark in text, the apparatus comprising: the extraction module is used for extracting the emotional characteristic mask of the text to be processed; the obtaining module is used for obtaining a ciphertext watermark of the text to be processed under the condition that the watermark identification exists in the text to be processed, wherein the ciphertext watermark is formed by embedding the emotional characteristic mask code into the text to be processed according to a conversion rule; the determining module is used for determining the matching degree between the emotional characteristic mask and the ciphertext watermark; and the detection module is used for carrying out ciphertext watermark detection on the text to be processed according to the matching degree.
In a third aspect, the present application provides a computer device, which includes a memory and a processor, where the memory stores a computer program executable on the processor, and the processor implements the text watermark embedding and detecting method when executing the program.
In a fourth aspect, the present application provides a storage medium for detecting a watermark in a text, and the storage medium stores executable instructions for causing a processor to implement the text watermark embedding and detecting method.
The application provides a method, a device, equipment and a storage medium for detecting a watermark in a text. Therefore, firstly, the characteristics of the text to be processed are fully considered, the text emotional characteristics capable of representing the text content are subjected to mask processing and are implicitly embedded into the text, the emotional characteristics are selected to well reflect the content tendency of the text, and the text content is represented. And secondly, by using a watermark embedding method based on the text emotional characteristics, the text characteristics after mask processing are small in size and low in calculation cost, the processing efficiency is improved, the carrying amount of redundant data is reduced, and the transparency and the safety of the text watermark are guaranteed by using a processing method for separating text content from the text watermark. And finally, determining the matching degree by using the emotional characteristic mask and the ciphertext watermark to realize ciphertext watermark detection on the text to be processed, wherein the text watermark still can play a role under the condition of certain interference or attack, and the robustness of the text watermark is improved.
Drawings
Fig. 1 is a schematic flowchart of an implementation process for detecting a watermark in a text according to an embodiment of the present disclosure;
FIG. 2A is a schematic diagram of an improved text emotion model based on a long-short term memory network model according to an embodiment of the present application;
fig. 2B is a schematic flowchart of an implementation process for detecting a watermark in a text according to an embodiment of the present application;
fig. 3 is a schematic flowchart of an implementation process for detecting a watermark in a text according to an embodiment of the present application;
fig. 4 is a schematic flowchart of an implementation process for detecting a watermark in a text according to an embodiment of the present application;
fig. 5 is a schematic structural diagram illustrating a structure of an apparatus for detecting a watermark in a text according to an embodiment of the present application;
fig. 6 is a hardware entity diagram of a computer device according to an embodiment of the present disclosure.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application.
It should be understood that some of the embodiments described herein are only for explaining the technical solutions of the present application, and are not intended to limit the technical scope of the present application.
As shown in fig. 1, a method for detecting a watermark in a text provided by an embodiment of the present application includes:
s101, extracting an emotional feature mask of a text to be processed;
the emotional characteristics refer to characteristics capable of reflecting content trends of the text, and the selected emotional characteristics can represent the content of the text. The mask may be a string of binary codes. Here, the emotional feature mask may be a string of binary codes capable of characterizing the text content tendency characteristics.
Step S102, under the condition that the watermark identification exists in the text to be processed, a ciphertext watermark of the text to be processed is obtained, wherein the ciphertext watermark is formed by embedding the emotional characteristic mask code into the text to be processed according to a conversion rule;
here, the watermark identifier may be embedded in the text according to actual needs, and is used to distinguish the text in which the watermark is embedded from the text in which the watermark is not embedded. For example: the content identified by the watermark may be logo information of the creator and owner of the text. And in the case of determining that the watermark identification exists in the text to be processed, namely determining that the text to be processed is a file in which the watermark embedding is finished. The watermark embedded file can embed three types of watermarks: the first kind, embedding the plain text watermark, the main purpose is to clarify the copyright, prevent the illegal use, the plain text watermark content can be self-defined according to the need; the second type is that a ciphertext watermark is embedded to prevent the file from being damaged or tampered; and thirdly, embedding watermark identification for protecting the rights and interests of legal owners of the texts.
The ciphertext watermark is formed by embedding the emotional characteristic mask into the text according to the conversion rule. Here, the conversion rule may be a single code (Unicode) conversion rule. Because computers can only process numbers, if text is to be processed, it must first be converted to numbers. Unicode is a character encoding scheme established by the international organization that can accommodate all the words and symbols in the world.
Step S103, determining the matching degree between the emotional characteristic mask and the ciphertext watermark;
because the ciphertext watermark is formed by embedding the emotional characteristic mask into the text to be processed according to the conversion rule, the matching value between the emotional characteristic mask and the ciphertext watermark can be obtained through calculation. The ciphertext watermark may be a different ciphertext watermark generated for different content of each page of text. When the text to be processed is a single page text, the matching degree refers to a matching value; when the text to be processed is a multi-page text, the matching degree is a set of matching value sets, wherein the number of the matching value sets is the same as the page number of the text to be processed.
And S104, carrying out ciphertext watermark detection on the text to be processed according to the matching degree.
And determining whether the text to be processed is the text in which the ciphertext watermark is embedded or determining whether the ciphertext watermark of the text to be processed is damaged or tampered in the transmission process according to the matching degree.
In the embodiment of the application, firstly, the emotional feature mask of the text to be processed is extracted, secondly, the ciphertext watermark is obtained under the condition that the watermark identification exists in the text to be processed, then, the matching degree between the emotional feature mask and the ciphertext watermark is determined, and finally, ciphertext watermark detection is carried out on the text to be processed according to the matching degree. Therefore, firstly, the characteristics of the text to be processed are fully considered, the text emotional characteristics capable of representing the text content are subjected to mask processing and are implicitly embedded into the text, the selected emotional characteristics can well reflect the content tendency of the text, and the text content is represented. And secondly, by using a watermark embedding method based on the text emotional characteristics, the text characteristics after mask processing are small in size and low in calculation cost, the processing efficiency is improved, the carrying amount of redundant data is reduced, and the transparency and the safety of the text watermark are guaranteed by using a processing method for separating text content from the text watermark. And finally, determining the matching degree by using the emotional characteristic mask and the ciphertext watermark to realize ciphertext watermark detection on the text to be processed, wherein the text watermark still can play a role under the condition of certain interference or attack, and the robustness of the text watermark is improved.
Fig. 2A is a schematic diagram of an improved text emotion model based on a long-short term memory network model according to an embodiment of the present application, as shown in fig. 2A, including: word vector 201, emotion vector 202, sentence emotion vector 203, Long Short-Term Memory network (LSTM) 204, normalization layer (Softmax)205, full connectivity layer (FC)206, and emotional characteristics (f)207, wherein,
the word vector 201, which is used to analyze each word in the sentence through the text structuring model and input as word embedding, the word embedding in a sentence can be expressed as { w }1,w2,…,wiAnd one word corresponds to one word vector.
An emotion vector 202 for embedding emotion into an input emotion tendency information to be determined, { v }1,v2,…,viV. Emotion embedding corresponding to all words, vi＝{p1,…,pMM is the number of emotion polarities to be judged, usually 3, corresponding to emotion polarity { positive, negative, neutral }, pMThe probability corresponding to each emotion polarity is 1 in total;
first feature vector w after connecting word vector 201 and emotion vector 202Tv is input into a long-short term memory network (LSTM) to extract the current stage of each sentence, and a second feature vector is obtained.
First feature vector w after connecting word vector 201 and emotion vector 202Tv is normalized (Softmax)205 to obtain a sentence emotion vector 203, h ═ h1,h2,…,hiAnd (3) extracting sentence emotion of the current stage of each sentence by using the following formula (1) to obtain a sentence emotion vector 203.
hi＝Softmax(wi Tvi) (1)；
The last layer of the long-short term memory network 204 is used for adding the second feature vector of the last stage and the sentence emotion vector 203 of the full stage, h ═ { h { (h) }1,h2,…,hiAnd (5) performing linear transformation on the output through a full connection layer FC, so that the vector length of the output is the same as the emotion polarity to be judged.
The output of the model is taken as the emotional feature with the highest probability by a normalization (Softmax) layer 205, and the probability of the output belonging to each category is expressed by the following formula (2).
f＝[f1,f2,…,fM] (2)；
Wherein M is the number of emotion polarities to be judged.
The improved text emotion model based on the long-short term memory network model shown in fig. 2A is input into the long-short term memory network model based on the target text represented by the fusion vector, and the emotion type of the target text is determined.
Fig. 2B is a method for detecting a watermark in a text according to an embodiment of the present application, and as shown in fig. 2B, the method includes:
step S201, converting the acquired first feature vector into a second feature vector by using a long-short term memory network, wherein the first feature vector comprises a sample word vector and an emotional tendency vector;
as shown in fig. 2A, each word in a sentence is converted into a corresponding word vector 201; the emotional tendency information to be determined is converted into corresponding emotional tendency vectors 202. Here, a large amount of sample words and emotional tendency information are input. Sample words and emotional tendency information can be correspondingly acquired according to the actual needs of the text to be processed.
Analyzing each word in the sentence through the text structured model and inputting the word as the word input in one sentenceMay be expressed as w1,w2,…,wiAnd one word corresponds to one word vector.
{v1,v2,…,viV. Emotion embedding corresponding to all words, vi＝{p1,…,pMM is the number of emotion polarities to be determined, for example, 3 may be selected, corresponding to the emotion polarities { positive, negative, neutral }, pMThe sum is 1 for the probability corresponding to each emotion polarity.
Connecting the word vector 201 and the emotion vector 202 to obtain a first feature vector wTv。
The Long Short-Term Memory Network (LSTM) is a time-cycle Neural Network, which is specially designed for solving the Long-Term dependence problem of the general cycle Neural Network (RNN), and all RNNs have a chain form of a repeated Neural Network module. Due to the unique design structure, LSTM is suitable for handling and predicting significant events in time series that are very long-spaced and delayed. Word information of emotional tendency and sentence emotional information of dynamic change are embedded into the long-term and short-term memory network training process, so that the long-term and short-term memory can select the currently most suitable emotional polarity through context information.
As shown in fig. 2A, a first feature vector w obtained by connecting a word vector 201 and an emotion vector 202Tv, inputting the current stage of each sentence into a long-short term memory network (LSTM) to obtain a second feature vector.
Step S202, normalization processing is carried out on the first feature vector by utilizing a normalization layer to obtain a sentence emotion vector;
as shown in fig. 2A, the first feature vector wTv is normalized (Softmax)205 to obtain a sentence emotion vector 203, h ═ h1,h2,…,hiAnd (3) extracting sentence emotion of the current stage of each sentence by using the following formula (1) to obtain a sentence emotion vector 203.
hi＝Softmax(wi Tvi) (1)。
Step S203, performing linear transformation on the second feature vector and the sentence emotion vector by using a full connection layer to obtain a third feature vector, so that the dimension number of the third feature vector is the same as the number of emotion polarities needing to be judged;
fully connected layers (FC) function as "classifiers" throughout the network, i.e., integrate highly abstracted features previously convolved multiple times. And performing linear transformation on the last layer output of the long-short term memory network through a full connection layer, so that the dimension number of the second feature vector is the same as the emotion polarity number required to be judged.
As shown in fig. 2A, the last stage of the long-short term memory network 204 is divided into the second feature vector of the last stage and the sentence emotion vector 203 of the full stage, h ═ { h { (h) } in the last layer1,h2,…,hiAnd (5) performing linear transformation on the output through a full connection layer FC, so that the vector length of the output is the same as the emotion polarity required to be judged.
Step S204, normalization processing is carried out on the third feature vector by utilizing a normalization layer, and pre-training of the text emotion model is completed;
here, the second feature vector may be normalized using a normalization (softmax) layer, and the pre-training of the text emotion model may be completed. The softmax layer is generally the last layer of the classification network, takes a feature vector with the same length and the same number of categories as output, and then outputs the probability that the vector formed by connecting the words and the emotions belongs to each category.
As shown in fig. 2A, the emotional features with the highest probability are taken as the output of the model by a normalization (Softmax) layer 205, and the probability of the output belonging to each category is expressed by the following formula (2).
f＝[f1,f2,…,fM] (2)；
Wherein M is the number of emotion polarities to be judged.
In this way, pre-training of the text emotion model can be achieved.
S205, extracting the emotional characteristics of the text to be processed by using the text emotional model;
here, the emotion characteristics of the text to be processed can be extracted by using the pre-trained text emotion model.
Step S206, carrying out binarization processing on the emotional characteristics to obtain an emotional characteristic mask;
and in consideration of high calculation cost of the feature data, performing binarization processing on the acquired text emotion features to obtain a text emotion feature mask convenient to calculate.
Step S207, under the condition that the watermark identification exists in the text to be processed, a ciphertext watermark of the text to be processed is obtained, wherein the ciphertext watermark is formed by embedding the emotional characteristic mask code into the text to be processed according to a conversion rule;
step S208, determining the matching degree between the emotional characteristic mask and the ciphertext watermark;
and S209, performing ciphertext watermark detection on the text to be processed according to the matching degree.
In the embodiment of the application, the text emotional feature extraction method based on the long and short term memory network combines word embedding and emotion embedding, and utilizes the improved long and short term memory network to perform fine-grained emotional analysis and extract the text emotional feature; the watermark embedding method based on the text emotional characteristics performs mask processing on the text emotional characteristics capable of representing the text content and embeds the text emotional characteristics into the text in an implicit mode. Therefore, the extraction precision of the text emotion model is improved from the two aspects of word embedding and emotion embedding, the text emotion characteristics are extracted by using the improved text emotion model based on the long-term and short-term memory network, even if various emotion tendencies exist in the text at the same time, the text emotion model can also extract the real emotion characteristics of the text, and the accuracy and the robustness of the text watermark are improved; by using the watermark embedding method based on the text emotional characteristics, the text characteristics after mask processing are small in size and low in calculation cost, the processing efficiency is improved, the carrying amount of redundant data is reduced, and the transparency and the safety of the text watermark are guaranteed by the processing method for separating the text content from the text watermark.
The embodiment of the application provides a method for embedding a watermark into a text without adding the watermark, which comprises the following steps:
step S211, acquiring a plaintext watermark of the text to be processed;
the main purpose of embedding the plain text watermark is to clarify copyright and prevent illegal use. The plain text watermark content can be self-defined according to actual needs.
Step S212, under the condition that the to-be-processed text is determined to have no watermark identification, embedding the plaintext watermark into the to-be-processed text;
in the step, the watermark algorithm based on discrete wavelet change can be adopted to complete plaintext watermark embedding.
Step S213, converting the emotional feature mask into a first zero-width control field according to a single code conversion rule;
unicode (Unicode), also known as Unicode and universal code, is an industry standard in the field of computer science, and includes character sets, encoding schemes, and the like. Unicode is generated to solve the limitation of the traditional character encoding scheme, and sets a uniform and unique binary code for each character in each language so as to meet the requirements of cross-language and cross-platform text conversion and processing. Because computers can only process numbers, if text is to be processed, it must first be converted to numbers.
The Unicode zero-width control field is a special Unicode field, is invisible and unprintable control field, is present in a page and is mainly used for adjusting the display format of characters, and the method is simple to implement and has small time and space loss.
Here, since the emotion feature mask is composed of 0 and 1, the emotion feature mask is converted into a corresponding Unicode code, resulting in a first zero-width control field.
Step S214, embedding the first zero-width control field into a first specific position of the text to be processed;
the first specific location may be the end of each page of text to be processed. Here, a first zero-width control field is embedded in the end of each page of the text to be processed to form an invisible ciphertext watermark.
Step S215, acquiring a watermark identifier of the text to be processed;
the watermark identification (Flag) content of the text to be processed can be the Flag information of the creator and the owner of the text, so as to protect the rights and interests of the legal owner of the text.
Step S216, converting the watermark identification into a second zero width control field according to a single code conversion rule;
and S217, embedding the second zero-width control field into a second specific position of the text to be processed, and outputting the text embedded with the watermark.
The first specific location may be a header of each page of the text to be processed. Here, a second zero-width control field is embedded in the tail of each page of the text to be processed to form watermark identification of the mark information of the creator and the owner of the text.
In the embodiment of the application, a plaintext watermark is embedded into a text to be processed, an emotional characteristic mask is embedded into the text to be processed as a ciphertext watermark, and a watermark identifier is embedded into the text to be processed, so that the watermark is added to the text without the watermark. Therefore, three types of watermarks with different functions are added to the text without water, and the rights and interests of legal text owners can be effectively protected.
Fig. 3 is a method for detecting a watermark in a text, where the text to be processed at least includes a page of text, and the matching degree is a set of matching values of the text to be processed, as shown in fig. 3, the method includes:
s301, extracting an emotional feature mask of the text to be processed;
inputting the ith page of text into the text emotion model in the step S403 to obtain the text emotion feature f of the ith page of texti；
And (3) traversing all pages of the text to obtain the text emotional characteristics of the whole text, wherein the following formula (3) represents the text emotional characteristic set of the whole text.
F＝{f1,f2,…,fi,…,fn} (3)；
Where n represents the total number of pages of text.
Considering that the feature data is high in calculation cost, the text emotion feature is subjected to binarization processing to obtain a text emotion feature mask convenient to calculate, and a text emotion feature mask set of the whole text is represented by a formula (4) below.
Mask＝{mask1,mask2,…,maski,…,maskn} (4)；
Wherein, maskiAnd the text emotion feature mask of the ith page of the text is represented.
Step S302, under the condition that the watermark identification exists in the text to be processed, a ciphertext watermark of the text to be processed is obtained, wherein the ciphertext watermark is formed by embedding the emotional characteristic mask code into the text to be processed according to a conversion rule;
step S303, determining a matching value of the emotional feature mask of each page of the text to be processed and the ciphertext watermark of each page of the text to be processed by utilizing a quadratic formula of a two-norm to obtain the matching value of each page of the text to be processed;
calculating the text emotion feature mask of the ith page of the textiAnd ciphertext watermark markiMatch value matchiThe following equation (5) is a square equation of the two-norm of the calculated matching value.
Wherein, matchiMatch value for i page cipher text watermark of textiThe larger the content of the text of the ith page is, the more similar the emotional characteristics reflected by the ciphertext watermark are.
Step S304, determining the matching value of each page of the text to be processed as the matching value in the matching value set of the text to be processed;
all pages of the text are traversed to obtain a ciphertext watermark matching value set of the whole text:
Match＝{match1,match2,…,matchi,…,matchn}。
step S305, under the condition that any matching value in the matching value set is larger than a first threshold value, determining that the text to be processed is embedded with a ciphertext watermark;
setting matchdIs a first threshold value, matchdAn empirical value of 0.05 can be taken;
when match existsi≥matchdThat is, the matching value of the ciphertext watermark of a certain page in the text is greater than the first threshold, which indicates that the text is embedded with the ciphertext watermark. And (4) circulating the number of the text pages, and comparing the relation between the ciphertext watermark matching value of each page and the watermark identification interference threshold value so as to judge whether the text is embedded with the ciphertext watermark.
When there is no matchi≥matchdWhen the watermark is input, the matching value of the ciphertext watermarks of all pages in the text is smaller than the watermark identification interference threshold, which indicates that the text is not embedded with the text watermark, the watermark identification may be interfered by channel noise, resampling, lossy compression and the like in transmission, the watermark identification of the input text is wrongly marked, and the watermark needs to be embedded into the text to be processed again.
Step S306, determining that the text to be processed passes through text watermark detection when each matching value in the matching value set is greater than the second threshold value, wherein the second threshold value is greater than the first threshold value.
Setting matchtAs a second threshold, matchtAn empirical value of 0.8 may be taken;
when matchi≥matchtAnd when the matching value of the ciphertext watermarks of all the pages in the text is larger than the second threshold value, the text is transmitted safely, namely the text is embedded with the ciphertext watermarks.
Step S307, under the condition that any matching value in the matching value set is smaller than the second threshold value, determining that the text to be processed is a text which does not pass ciphertext watermark detection;
when match existsi＜matchtWhen the matching value of the ciphertext watermark of a certain page in the text is smaller than the ciphertext watermark tampering threshold value, the text is possibly tampered in the transmission process,and determining that the text to be processed is the text which does not pass the ciphertext watermark detection.
And S308, issuing a tamper early warning to the text which does not pass the ciphertext watermark detection so as to prompt that the text to be processed has a tamper risk.
In the embodiment of the application, a multiple threshold screening mechanism is used, under certain interference or attack, the text watermark model can correctly complete watermark embedding or detection functions, and can send tamper early warning to texts with tamper tendencies. Therefore, under the condition of being interfered or attacked to a certain degree, the text watermark can still play a role, and the robustness of the text watermark is improved.
Fig. 4 is a method for detecting a watermark in a text according to an embodiment of the present application, and as shown in fig. 4, the method includes:
step S401, inputting a text to be processed;
the text to be processed needs to have page attributes, such as pdf, doc and other types of text;
step S402, pre-training a text emotion model;
in order to improve the stability and efficiency of feature calculation, the feature model needs to be pre-trained in the step. In consideration of the characteristics of text data, the step improves a text emotion model on the basis of a long-term and short-term memory network model, and the model is designed as shown in FIG. 2A.
Inputting each word in the sentence as a word vector 201; inputting emotion tendency information to be judged as emotion embedding vectors 202; connecting the word vector 201 and the emotion vector 202 to obtain a first feature vector wTv; inputting the first feature vector into a long-short term memory network to extract emotional features, and obtaining a second feature vector; normalizing the first feature vector to obtain a sentence emotion vector 203; inputting the second feature vector and the sentence emotion vector into a full-connection layer for linear transformation to obtain a third feature vector, so that the dimension number of the third feature vector is the same as the emotion polarity number required to be judged; finally, the emotional feature f with the highest probability is taken as the output of the model through a normalization (Softmax) layer 205.
Thus, the target text expressed based on the fusion vector can be input into the long-term and short-term memory network model together with the trained emotional characteristic model, and the emotion type of the target text can be determined.
Step S403, calculating text emotional characteristics;
acquiring text emotional characteristics, inputting the ith page of text into the text emotional model in the step S403 to obtain text emotional characteristics f of the ith page of texti。
And (3) traversing all pages of the text to obtain the text emotional characteristics of the whole text, wherein the following formula (3) represents the text emotional characteristic set of the whole text.
F＝{f1,f2,…,fi,…,fn} (3)；
Where n represents the total number of pages of the text.
Step S404, generating a text emotion feature mask;
considering that the feature data has high calculation cost, the text emotion feature acquired in step S403 is subjected to binarization processing to obtain a text emotion feature mask convenient for calculation, and the following formula (4) represents a text emotion feature mask set of the whole text.
Mask＝{mask1,mask2,…,maski,…,maskn} (4)；
Wherein, maskiAnd the text emotion feature mask of the ith page of the text is represented.
Step S405, detecting whether a watermark identification exists;
decoding a Unicode zero-width control field at the head of the text according to a conversion rule to obtain a text watermark identification Flag;
when Flag is not empty, it indicates that the text watermark is embedded, go to step S405 to enter a watermark detection process;
and when the Flag is empty, the text is indicated to be not embedded with the text watermark, and a watermark embedding process is carried out.
Step S406, embedding a plain text watermark;
the main purpose of embedding the plain text watermark is to clarify copyright and prevent illegal use. The plain text watermark content can be self-defined, and no restriction is imposed here. In the step, the watermark algorithm based on discrete wavelet change can be adopted to complete plaintext watermark embedding.
Step S407, embedding the ciphertext watermark;
will maskiReplacing the conversion rule with a Unicode zero-width control field and embedding the control field into the tail of each page to form an invisible ciphertext watermark; traversing all pages of the text, and embedding a ciphertext watermark into each page;
step S408, embedding a watermark mark;
setting a text watermark identification Flag, wherein the Flag content is the Flag information of a creator and an owner of the text and is used for protecting the rights and interests of a legal owner of the text; the Flag is replaced with a Unicode zero width control field embedded in the text header.
Step S409, outputting a watermark text;
after the above steps S405, S406, and S407, watermark embedding is completed, and a watermark text is output.
Step S410, ciphertext watermarks are matched to obtain a matching value;
decoding the Unicode zero-width control field at the tail part of each page according to the conversion rule to obtain the ciphertext watermark mark of the ith page of the texti；
Calculating the text emotion feature mask of the ith page of the textiAnd ciphertext watermark markiMatch value matchiThe following formula (5) is a formula for calculating the matching value.
Wherein, matchiMatch value for i page cipher text watermark of textiThe larger the content is, the more similar the emotional characteristics reflected by the text content of the ith page and the ciphertext watermark are;
all pages of the text are traversed to obtain a ciphertext watermark matching value set of the whole text:
Match＝{match1,match2,…,matchi,…,matchn}。
step S411, determining whether the matching value is larger than an interference threshold value;
determining whether the matching value is greater than an interference threshold (first threshold), and setting matchdIdentifying an interference threshold for a watermark, matchdTaking an empirical value of 0.05;
when match existsi≥matchdIf the matching value of the ciphertext watermark of a certain page in the text is larger than the watermark identification interference threshold, indicating that the text is embedded with the text watermark, turning to step 503;
when there is no matchi≥matchdWhen the text is input, namely ciphertext watermark matching values of all pages in the text are smaller than a watermark identification interference threshold, the text is indicated to be not embedded with the text watermark, the watermark identification is possibly interfered by channel noise, resampling, lossy compression and the like in transmission, the watermark identification of the input text is wrongly marked, and the watermark needs to be embedded again;
step S412, determining whether the matching value is larger than a tampering threshold value (second threshold value);
setting matchtTampering with threshold for ciphertext watermarks, matchtTaking an empirical value of 0.8;
when matchi≥matchtWhen the result is positive, the ciphertext watermark matching values of all pages in the text are larger than the ciphertext watermark tampering threshold value, and the text is transmitted safely;
when match existsi＜matchtWhen the text is in a transmission state, namely, the ciphertext watermark matching value of a certain page in the text is smaller than the ciphertext watermark tampering threshold value, the text is possibly tampered in the transmission process;
step S413, outputting a text to be processed;
judging in step S412 that the input text to be processed is not tampered, and directly outputting the input text through watermark detection;
step S414, issuing text tampering early warning;
and S412 judges that the inputted text to be processed has tampering risk, and the text tampering early warning is issued without passing watermark detection.
The embodiment of the application provides a text watermark model based on emotional characteristics, a text to be processed is input, the text emotional characteristics can be extracted by using the text emotional model, and a watermark text is generated by embedding the text; the watermark embedded text can automatically detect the watermark and send a tamper early warning to the text with the tamper risk. The text emotional feature extraction method based on the long and short term memory network combines word embedding and emotion embedding, and utilizes the improved long and short term memory network to perform fine-grained emotional analysis and extract text emotional features; the watermark embedding method based on the text emotional characteristics comprises the steps of performing mask processing on the text emotional characteristics capable of representing text content and embedding the text emotional characteristics into a text in an implicit mode; and the multiple threshold screening mechanism can correctly complete the watermark embedding or detecting function of the text watermark model under certain interference or attack, and can send tamper early warning to the text with tamper tendency. In this way, the improved text emotion model based on the long-term and short-term memory network is used for extracting the text emotion characteristics, so that the text emotion model can also extract the real emotion characteristics of the text even if various emotion tendencies exist in the text at the same time, and the accuracy and the robustness of the text watermark are improved; by using the watermark embedding method based on the text emotional characteristics, the text characteristics after mask processing are small in size and low in calculation cost, the processing efficiency is improved, the carrying amount of redundant data is reduced, and the transparency and the safety of the text watermark are guaranteed by using the processing method for separating the text content from the text watermark; by using a multiple threshold screening mechanism, the text watermark can still play a role under the condition of being interfered or attacked to a certain degree, and the robustness of the text watermark is improved.
Based on the foregoing embodiments, the present application provides an apparatus for detecting a watermark in a text, where the apparatus includes modules that can be implemented by a processor in a computer device; of course, the implementation can also be realized through a specific logic circuit; in implementation, the processor may be a Central Processing Unit (CPU), a Microprocessor (MPU), a Digital Signal Processor (DSP), a Field Programmable Gate Array (FPGA), or the like.
Fig. 5 is a schematic structural diagram of a composition of a text watermark embedding and detecting device according to an embodiment of the present application, and as shown in fig. 5, the device 500 includes an extracting module 501, a first obtaining module 502, a determining module 503, and a detecting module 504, where:
an extraction module 501, configured to extract an emotional feature mask of a text to be processed;
a first obtaining module 502, configured to obtain a ciphertext watermark of the to-be-processed text when it is determined that the to-be-processed text has a watermark identifier, where the ciphertext watermark is formed by embedding the emotional feature mask in the to-be-processed text according to a conversion rule;
a determining module 503, configured to determine a matching degree between the emotional feature mask and the ciphertext watermark;
and the detection module 504 is configured to perform ciphertext watermark detection on the text to be processed according to the matching degree.
Based on the foregoing embodiment, the extraction module includes an extraction submodule and a binarization processing module, wherein the extraction submodule is configured to extract, by using a text emotion model, emotion features of the text to be processed; and the binarization processing module is used for carrying out binarization processing on the emotional features to obtain the emotional feature mask.
Based on the foregoing embodiment, the extraction module further includes an obtaining conversion sub-module, a first normalization processing sub-module, a linear transformation module, and a second normalization processing sub-module, where the conversion sub-module is configured to convert the obtained first feature vector into a second feature vector by using a long-short term memory network, where the first feature vector includes sample words and emotional tendency information; the first normalization submodule is used for performing normalization processing on the first feature vector to obtain a sentence emotion vector; the linear transformation submodule is used for inputting the second eigenvector and the sentence emotion vector into a full connection layer for linear transformation to obtain a third eigenvector, so that the dimension number of the third eigenvector is the same as the emotion polarity number required to be judged; and the second normalization processing submodule is used for performing normalization processing on the third feature vector to finish the pre-training of the text emotion model.
Based on the foregoing embodiment, the apparatus further includes a second obtaining module, a first embedding module, a first converting module, a second embedding module, a third obtaining module, a second converting module, and a third embedding module, where the second obtaining module is configured to obtain a plaintext watermark of the to-be-processed text; the first embedding module is used for embedding the plaintext watermark into the text to be processed under the condition that the watermark identification does not exist in the text to be processed; the first conversion module is used for converting the emotional feature mask into a first zero-width control field according to a single code conversion rule; the second embedding module is used for embedding the first zero-width control field into a first specific position of the text to be processed; the third obtaining module is configured to obtain a watermark identifier of the text to be processed; the second conversion module is used for converting the watermark identifier into a second zero-width control field according to a single code conversion rule; and the third embedding module is used for embedding the second zero-width control field into a second specific position of the text to be processed and outputting the text embedded with the watermark.
Based on the foregoing embodiment, the to-be-processed text at least includes a page of text, the matching degree is a set of matching values of the to-be-processed text, and the determining module includes: the first determining submodule is used for determining an emotional feature mask of each page of the text to be processed and a matching value of a ciphertext watermark of each page of the text to be processed by utilizing a square formula of a two-norm to obtain the matching value of each page of the text to be processed; the second determining sub-module is configured to determine a matching value of each page of the text to be processed as a matching value in the set of matching values of the text to be processed.
Based on the foregoing embodiment, the detection module includes a third determining sub-module, a fourth determining sub-module, a fifth determining sub-module, and a tamper early warning module, where the third determining sub-module is configured to determine that the text to be processed is embedded with a ciphertext watermark if any matching value in the matching value set is greater than a first threshold value; the fourth determining sub-module is configured to determine that the text to be processed passes through text watermark detection if each matching value in the set of matching values is greater than the second threshold, where the second threshold is greater than the first threshold; the fifth determining submodule is configured to determine that the text to be processed is a text that does not pass ciphertext watermark detection when any matching value in the matching value sets is smaller than the second threshold; the tampering early warning module is used for issuing tampering early warning to the text which does not pass ciphertext watermark detection so as to prompt that the text to be processed has tampering risks.
The above description of the apparatus embodiments, similar to the above description of the method embodiments, has similar beneficial effects as the method embodiments. For technical details not disclosed in the embodiments of the apparatus of the present application, reference is made to the description of the embodiments of the method of the present application for understanding.
It should be noted that, in the embodiment of the present application, if the text watermark embedding and detecting method is implemented in the form of a software functional module, and is sold or used as a standalone product, it may also be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the embodiments of the present application or portions thereof that contribute to the related art may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling an electronic device (which may be a mobile phone, a tablet computer, a notebook computer, a desktop computer, etc.) to execute all or part of the methods described in the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read Only Memory (ROM), a magnetic disk, or an optical disk. Thus, embodiments of the present application are not limited to any specific combination of hardware and software.
Correspondingly, the present application provides a computer-readable storage medium, on which a computer program is stored, and the computer program, when executed by a processor, implements the steps in the method for detecting a watermark in a text provided in the above embodiments.
Correspondingly, an embodiment of the present application provides a computer device, fig. 6 is a schematic diagram of a hardware entity of the computer device in the embodiment of the present application, and as shown in fig. 6, the hardware entity of the device 600 includes: comprising a memory 601 and a processor 602, said memory 601 storing a computer program operable on the processor 602, said processor 602 implementing the steps of detecting a watermark in text as provided in the above embodiments when executing said program.
The Memory 601 is configured to store instructions and applications executable by the processor 602, and may also buffer data (e.g., image data, audio data, voice communication data, and video communication data) to be processed or already processed by the processor 602 and modules in the computer device 600, and may be implemented by a FLASH Memory (FLASH) or a Random Access Memory (RAM).
Here, it should be noted that: the above description of the storage medium and device embodiments is similar to the description of the method embodiments above, with similar advantageous effects as the method embodiments. For technical details not disclosed in the embodiments of the storage medium and the apparatus of the present application, reference is made to the description of the embodiments of the method of the present application for understanding.
It should be appreciated that reference throughout this specification to "one embodiment" or "an embodiment" means that a particular feature, structure or characteristic described in connection with the embodiment is included in at least one embodiment of the present application. Thus, the appearances of the phrases "in one embodiment" or "in an embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. It should be understood that, in the various embodiments of the present application, the sequence numbers of the above-mentioned processes do not mean the execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present application. The above-mentioned serial numbers of the embodiments of the present application are merely for description and do not represent the merits of the embodiments.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. The above-described device embodiments are merely illustrative, for example, the division of the unit is only a logical functional division, and there may be other division ways in actual implementation, such as: multiple units or components may be combined, or may be integrated into another system, or some features may be omitted, or not implemented. In addition, the coupling, direct coupling or communication connection between the components shown or discussed may be through some interfaces, and the indirect coupling or communication connection between the devices or units may be electrical, mechanical or in other forms.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units; can be located in one place or distributed on a plurality of network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, all functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may be separately regarded as one unit, or two or more units may be integrated into one unit; the integrated unit can be realized in a form of hardware, or in a form of hardware plus a software functional unit.
Those of ordinary skill in the art will understand that: all or part of the steps for realizing the method embodiments can be completed by hardware related to program instructions, the program can be stored in a computer readable storage medium, and the program executes the steps comprising the method embodiments when executed; and the aforementioned storage medium includes: various media that can store program codes, such as a removable Memory device, a Read Only Memory (ROM), a magnetic disk, or an optical disk.
Alternatively, the integrated units described above in the present application may be stored in a computer-readable storage medium if they are implemented in the form of software functional modules and sold or used as independent products. Based on such understanding, the technical solutions of the embodiments of the present application may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a mobile phone, a tablet computer, a notebook computer, a desktop computer, etc.) to execute all or part of the methods described in the embodiments of the present application. And the aforementioned storage medium includes: various media that can store program code, such as removable storage devices, ROMs, magnetic or optical disks, etc.
The methods disclosed in the several method embodiments provided in the present application may be combined arbitrarily without conflict to obtain new method embodiments.
Features disclosed in several of the product embodiments provided in the present application may be combined in any combination to yield new product embodiments without conflict.
The features disclosed in the several method or apparatus embodiments provided in the present application may be combined arbitrarily, without conflict, to arrive at new method embodiments or apparatus embodiments.
The above description is only for the embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of changes or substitutions within the technical scope of the present application, and shall be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (10)
1. A method of detecting a watermark in text, the method comprising:
extracting an emotional characteristic mask of a text to be processed;
under the condition that the watermark identification exists in the text to be processed, acquiring a ciphertext watermark of the text to be processed, wherein the ciphertext watermark is formed by embedding the emotional characteristic mask code into the text to be processed according to a conversion rule;
determining the matching degree between the emotional characteristic mask and the ciphertext watermark;
and carrying out ciphertext watermark detection on the text to be processed according to the matching degree.
2. The method of claim 1, wherein extracting the emotion feature mask of the text to be processed comprises:
extracting the emotional characteristics of the text to be processed by using a text emotional model;
and carrying out binarization processing on the emotional features to obtain the emotional feature mask.
3. The method of claim 2, wherein the text emotion model is trained by:
converting the acquired first feature vector into a second feature vector by using a long-short term memory network; wherein the first feature vector comprises a sample word vector and an emotional tendency vector;
normalizing the first feature vector by using a normalization layer to obtain a sentence emotion vector;
performing linear transformation on the second feature vector and the sentence emotion vector by using a full connection layer to obtain a third feature vector, so that the dimension number of the third feature vector is the same as the emotion polarity number required to be judged;
and carrying out normalization processing on the third feature vector by utilizing the normalization layer to finish the pre-training of the text emotion model.
4. The method of any of claims 1 to 3, further comprising:
acquiring a plaintext watermark of the text to be processed;
embedding the plaintext watermark into the text to be processed under the condition that the watermark identification does not exist in the text to be processed;
converting the emotional feature mask into a first zero-width control field according to a single code conversion rule;
embedding the first zero-width control field in a first specific position of the text to be processed;
acquiring a watermark identifier of the text to be processed;
converting the watermark identification into a second zero-width control field according to a single code conversion rule;
and embedding the second zero-width control field into a second specific position of the text to be processed, and outputting the text embedded with the watermark.
5. The method according to any one of claims 1 to 3, wherein the text to be processed includes at least one page of text, the matching degree is a set of matching values of the text to be processed, and the determining the matching degree of the emotional feature mask and the ciphertext watermark includes:
determining an emotional characteristic mask of each page of the text to be processed and a matching value of a ciphertext watermark of each page of the text to be processed by utilizing a quadratic formula of a two-norm to obtain the matching value of each page of the text to be processed;
and determining the matching value of each page of the text to be processed as the matching value in the matching value set of the text to be processed.
6. The method of claim 5, wherein the ciphertext watermark detection of the text to be processed according to the matching degree comprises:
determining that the text to be processed is embedded into a ciphertext watermark if any one of the matching value sets is greater than a first threshold value;
determining that the text to be processed passes through text watermark detection if each matching value in the set of matching values is greater than the second threshold, wherein the second threshold is greater than the first threshold.
7. The method of claim 6, wherein the method further comprises:
determining that the text to be processed is a text which does not pass ciphertext watermark detection under the condition that any matching value in the matching value set is smaller than the second threshold value;
and issuing a tampering early warning to the text which does not pass the ciphertext watermark detection so as to prompt that the text to be processed has a tampering risk.
8. An apparatus for detecting a watermark in text, the apparatus comprising:
the extraction module is used for extracting the emotional feature mask of the text to be processed;
the obtaining module is used for obtaining a ciphertext watermark of the text to be processed under the condition that the watermark identification exists in the text to be processed, wherein the ciphertext watermark is formed by embedding the emotional characteristic mask code into the text to be processed according to a conversion rule;
the determining module is used for determining the matching degree between the emotional characteristic mask and the ciphertext watermark;
and the detection module is used for carrying out ciphertext watermark detection on the text to be processed according to the matching degree.
9. A computer device comprising a memory and a processor, the memory storing a computer program operable on the processor, wherein the processor implements the steps of the method of any one of claims 1 to 7 when executing the program.
10. A storage medium having stored thereon executable instructions for causing a processor to perform the steps of the method of any one of claims 1 to 7 when executed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011497653.7A CN114648435A (en) | 2020-12-17 | 2020-12-17 | Method, device and equipment for detecting watermark in text and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011497653.7A CN114648435A (en) | 2020-12-17 | 2020-12-17 | Method, device and equipment for detecting watermark in text and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114648435A true CN114648435A (en) | 2022-06-21 |
Family
ID=81989982
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011497653.7A Pending CN114648435A (en) | 2020-12-17 | 2020-12-17 | Method, device and equipment for detecting watermark in text and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114648435A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117272333A (en) * | 2022-10-28 | 2023-12-22 | 北京鸿鹄元数科技有限公司 | Relational database watermark embedding and tracing method |
| CN117272333B (en) * | 2022-10-28 | 2024-05-24 | 北京鸿鹄元数科技有限公司 | Relational database watermark embedding and tracing method |
-
2020
- 2020-12-17 CN CN202011497653.7A patent/CN114648435A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117272333A (en) * | 2022-10-28 | 2023-12-22 | 北京鸿鹄元数科技有限公司 | Relational database watermark embedding and tracing method |
| CN117272333B (en) * | 2022-10-28 | 2024-05-24 | 北京鸿鹄元数科技有限公司 | Relational database watermark embedding and tracing method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107330306B (en) | Text watermark embedding and extracting method and device, electronic equipment and storage medium | |
| CN110532381B (en) | Text vector acquisition method and device, computer equipment and storage medium | |
| Al-Wesabi | A Hybrid Intelligent Approach for Content Authentication and Tampering Detection of Arabic Text Transmitted via Internet. | |
| Xiang et al. | A word-embedding-based steganalysis method for linguistic steganography via synonym substitution | |
| Domain | A review and open issues of diverse text watermarking techniques in spatial domain | |
| Al-Wesabi | Proposing high-smart approach for content authentication and tampering detection of arabic text transmitted via internet | |
| Al-Wesabi | A smart English text zero-watermarking approach based on third-level order and word mechanism of Markov model | |
| Al-Wesabi | Entropy-Based Watermarking Approach for Sensitive Tamper Detection of Arabic Text. | |
| Memon et al. | EVALUATION OF STEGANOGRAPHY FOR URDU/ARABIC TEXT. | |
| Hilal et al. | A hybrid intelligent text watermarking and natural language processing approach for transferring and receiving an authentic english text via internet | |
| Al-Wesabi et al. | A Reliable NLP Scheme for English Text Watermarking Based on Contents Interrelationship. | |
| CN115130613A (en) | False news identification model construction method, false news identification method and device | |
| CN114580371A (en) | Program semantic confusion method and system based on natural language processing | |
| Han et al. | Text adversarial attacks and defenses: Issues, taxonomy, and perspectives | |
| Al-Wesabi et al. | Tampering Detection Approach of Arabic-Text Based on Contents Interrelationship. | |
| CN109670304A (en) | Recognition methods, device and the electronic equipment of malicious code family attribute | |
| CN117421423A (en) | Method for acquiring entity identification model, method and device for identifying entity | |
| CN114648435A (en) | Method, device and equipment for detecting watermark in text and storage medium | |
| CN107491423B (en) | Chinese document gene quantization and characterization method based on numerical value-character string mixed coding | |
| CN112507388B (en) | Word2vec model training method, device and system based on privacy protection | |
| Al-Wesabi et al. | Proposing a High-Robust Approach for Detecting the Tampering Attacks on English Text Transmitted via Internet. | |
| Alamgeer et al. | Smart-Fragile Authentication Scheme for Robust Detecting of Tampering Attacks on English Text. | |
| Al-Wesabi | Text Analysis-Based Watermarking Approach for Tampering Detection of English Text. | |
| CN113836297A (en) | Training method and device for text emotion analysis model | |
| Al Duhayyim et al. | Securing Arabic Contents Algorithm for Smart Detecting of Illegal Tampering Attacks. |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |