Vector signature method, device and system for preventing information leakage

Abstract

The present application relates to the field of data security technology, and specifically to a vector signature method, device, and system for preventing information leakage, the method comprising: obtaining all text data to be signed; obtaining the frequency feature value of each word based on the sensitivity feature and frequency feature of the words in all the text data to be signed; obtaining the semantic feature distance and frequency feature distance of each word based on the distance between each word and the corresponding word vector of other words in the text data to be signed and the difference in frequency feature value, and then mapping each word to a two-dimensional coordinate system; clustering the words in the two-dimensional coordinate system; selecting some words from all the text data to be signed for block division based on the semantic feature distance and frequency feature distance of all words in each cluster, and then generating a vector signature. The present application improves the security of vector signatures by deeply analyzing the semantic features and frequency features in the text data, selecting words that are easily tampered with, and then improving the security of vector signatures.

Description

Technical Field

[0001] The present application relates to the field of data security technology, and specifically to a vector signature method, device, and system for preventing information leakage. Background Art

[0002] The core value of vector signatures lies in establishing a mathematical binding relationship between content and signature. By converting electronic signatures into vector graphics with coordinate invariance and geometric uniqueness (such as official seals constructed from Bezier curves or handwritten signature paths), document integrity verification can be transformed into a "feature value comparison" process. If the text content is illegally tampered with (e.g., by removing confidential clauses, replacing key data fields, or adjusting core semantic logic), the text feature values ​​and the mapping information embedded in the signature will no longer match. The system can accurately identify information leakage risks by comparing feature value differences in real time. This technology plays a key role in scenarios with extremely high data security requirements, such as electronic contract signing. For example, in financial contracts, vector signatures can ensure the immutability of sensitive fields such as "amount," "repayment period," and "breach of contract liability clause" through feature value binding. This effectively prevents legal disputes and asset losses caused by signature forgery or text tampering, becoming a fundamental security barrier against information leakage in the digital age.

[0003] The currently widely used single-dimensional mechanism of "generating a signature from a text hash value" suffers from a significant lack of semantic relevance. Traditional hash calculations rely solely on bit-by-bit matching of character sequences, lacking a deep logical mapping to the text's semantics. This allows attackers to exploit the characteristics of natural language to launch "collision attacks," whereby attackers, through synonym substitution, construct tampered text with different semantics but the same character sequence hash value. In such attack scenarios, electronic signatures, unable to detect covert semantic manipulation, could lead to the illegal replacement or blocking of sensitive information, posing a serious risk of information leakage. Summary of the Invention

[0004] In order to solve the above technical problems, the purpose of this application is to provide a vector signature method, device and system to prevent information leakage. The technical solutions adopted are as follows:

[0005] In a first aspect, an embodiment of the present application provides a vector signature method for preventing information leakage, the method comprising the following steps:

[0006] Get all the text data to be signed;

[0007] Divide all text data to be signed into highly sensitive text and less sensitive text; obtain the word frequency statistical feature value of each word based on its word frequency in the text data to be signed and its word frequency in all text data to be signed; obtain the feature weight of the single-category sensitive text based on the ratio of the average level of the word frequency statistical feature values ​​of all words in the single-category sensitive text to the sum of the average levels of the word frequency statistical feature values ​​of all words in the two-category sensitive text; obtain the word frequency feature value of each word based on the word frequency statistical feature value of each word and the feature weight of the sensitive text of its corresponding category;

[0008] Based on the distance between the corresponding word vectors of each word and every other word in the text data to be signed and the difference in word frequency feature values, the semantic feature distance and word frequency feature distance of each word are obtained respectively, and used as the horizontal and vertical coordinates of each word to map each word to a two-dimensional coordinate system; based on the distance between any two words in the two-dimensional coordinate system and the distance between their corresponding word vectors, the words in the two-dimensional coordinate system are clustered, and all words are divided into multiple clusters; based on the average level of the semantic feature distance and the average level of the word frequency feature distance of all words in each cluster, some words are selected from all the text data to be signed for block division, and then a vector signature is generated.

[0009] Preferably, the process of obtaining the word frequency statistical feature value of each word is: taking the word frequency of each word in all the text data to be signed as the first feature value of each word; taking the word frequency of each word in the text data to be signed as the second feature value of each word; taking the average of the first feature value and the second feature value of each word as the word frequency statistical feature value of each word.

[0010] Preferably, the process of obtaining the feature weight of the single-category sensitive text is as follows: recording the mean of the word frequency statistical feature values ​​of all words in the low-sensitivity text as the first feature coefficient of the low-sensitivity text, and recording the mean of the word frequency statistical feature values ​​of all words in the high-sensitivity text as the second feature coefficient of the high-sensitivity text; calculating the sum of the first feature coefficient and the second feature coefficient, recording the ratio of the first feature coefficient to the sum as the feature weight of the low-sensitivity text, and recording the ratio of the second feature coefficient to the sum as the feature weight of the high-sensitivity text.

[0011] Preferably, the word frequency feature value of each word is the product of the word frequency statistical feature value of each word and the feature weight of the sensitive text of the corresponding category.

[0012] Preferably, the process of obtaining the semantic feature distance and word frequency feature distance of each word is as follows: the mean of the DTW distances of the corresponding word vectors between each word and each other word in the text data to be signed is used as the semantic feature distance of each word; the mean of the absolute differences in the word frequency feature values ​​between each word and each other word in the text data to be signed is used as the word frequency feature distance of each word.

[0013] Preferably, the specific process of dividing all words into multiple clusters is: obtaining the similarity substitution distance between any two words based on the distance between any two words in the two-dimensional coordinate system and the distance between their corresponding word vectors; using the coordinates of all words in the two-dimensional coordinate system as the input of the clustering algorithm, and using the similarity substitution distance between each word as the metric distance of the clustering algorithm to obtain the clustering results of all words; wherein the calculation formula of the similarity substitution distance between two words is: Where, Indicates the and The similarity substitution distance between words, Indicates the and The Euclidean distance between words, Indicates the and The DTW distance between the word vectors corresponding to the words.

[0014] Preferably, the specific process of selecting some words from all the text data to be signed and dividing them into blocks is:

[0015] Calculate the priority coefficient of each cluster: Where, represents the priority coefficient of the j-th cluster, represents the mean of the semantic feature distances of all words in the jth cluster, Represents the mean of the frequency feature distances of all words in the j-th cluster;

[0016] The order of cluster selection is determined according to the priority coefficients of all clusters from large to small. The order of vocabulary selection within the cluster is selected according to the product of the semantic feature distance and the word frequency feature distance corresponding to each vocabulary from large to small. The selected vocabulary is placed in a preset number of blocks in turn until the size of each block meets the preset fixed block size.

[0017] Preferably, the specific process of generating the vector signature is: performing iterative hash calculation on the words in the block after segmentation to generate a hash value; implicitly embedding the hash value into the vector signature template; using AES-256 symmetric encryption and user public key asymmetric encryption to encrypt the text data to be signed, binding the vector graphics embedded with the hash value to the encrypted text, and generating a visual vector signature through position anchoring and transparency control.

[0018] In a second aspect, an embodiment of the present application provides a vector signature device for preventing information leakage, wherein the load flexibility adjustment device includes: a data acquisition module, a semantic word frequency analysis module, and a vector signature generation module.

[0019] Data collection module, used to obtain all text data to be signed;

[0020] Semantic word frequency analysis module, used to obtain the word frequency feature value of each word based on the semantic sensitivity feature and word frequency feature of each word;

[0021] The vector signature generation module is used to divide each word into multiple clusters based on the semantic features and word frequency features of each word, and select some words for block division according to the priority of each cluster and the priority of each word in each cluster, and then generate vector signatures based on the words in the block after division.

[0022] In a third aspect, an embodiment of the present application also provides a vector signature system for preventing information leakage, the system comprising a memory, a processor, and a computer program stored in the memory and running on the processor, and when the processor executes the computer program, it implements the steps of any one of the above-mentioned vector signature methods for preventing information leakage.

[0023] This application has at least the following beneficial effects:

[0024] This application fully considers the semantic features of different texts in the text data to be signed, analyzes the word frequency features in different text data to be signed, constructs word frequency feature values, evaluates the impact of synonym replacement on the hash mapping results of text data, and highlights the sensitivity features of different word semantic replacements; based on the sensitivity feature analysis results, comprehensively considers the differences in semantic features and word frequency features between different texts in the text data block processing, adjusts the text data block in the hash mapping process, and gives priority to allocating more sensitive words to the blocks. Under the premise of round function calculation, if synonym replacement occurs, it will have a greater impact on the hash calculation results, thereby improving the sensitivity of tampering identification of the text data to be signed, enhancing the security of vector signatures, and reducing the risk of text leakage. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0026] Figure 1 A flowchart of a vector signature method for preventing information leakage provided in one embodiment of the present application;

[0027] Figure 2 A schematic diagram of the structure of a vector signature device for preventing information leakage provided in an embodiment of the present application. DETAILED DESCRIPTION

[0028] To further illustrate the technical means and effects adopted by this application to achieve the intended invention objectives, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features, and effects of a vector signature method, device, and system for preventing information leakage proposed in this application. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable form.

[0029] Unless otherwise specified and limited, terms such as "comprises", "includes" or any other variants thereof are intended to cover non-exclusive inclusion, so that a circuit structure, article or device comprising a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the article or device comprising the element. In addition, the term "and\or" used herein includes any and all combinations of one or more related listed items. All technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs.

[0030] The following describes in detail a specific scheme of a vector signature method, device and system for preventing information leakage provided by this application with reference to the accompanying drawings.

[0031] See also Figure 1 , which shows a flowchart of a vector signature method for preventing information leakage provided by an embodiment of the present application, the method comprising the following steps:

[0032] Step 1: Get all the text data to be signed.

[0033] In this application, the OA system and electronic signature platform are connected through API to obtain text data to be signed, including the company's contracts and approval documents, with a focus on texts that require handwritten signatures, official seals, and interline seals; it should be noted that the company's contracts and approval documents are obtained under authorization for the implementation and analysis of the method in this application.

[0034] Specifically, the LTP tool is used to perform Chinese word segmentation, part-of-speech tagging and named entity recognition, and word vectors are generated through Word2Vec; regular expressions are used to filter out garbled characters and repeated text, the character encoding is unified to UTF-8, and format parameters such as paragraph spacing are standardized. The SimHash algorithm is used to remove semantically repeated text, and finally the pre-processed word segmentation and tagging recognition results of each text data to be signed are obtained.

[0035] Step 2: Divide all texts in the text data to be signed into high-sensitivity texts and low-sensitivity texts; obtain the word frequency statistical feature value of each word based on its word frequency in the text data to be signed and its word frequency in all text data to be signed; obtain the feature weight of a single category of sensitive text based on the ratio of the average level of the word frequency statistical feature values ​​of all words in a single category of sensitive text to the sum of the average levels of the word frequency statistical feature values ​​of all words in the two categories of sensitive text; obtain the word frequency feature value of each word based on the word frequency statistical feature value of each word and the feature weight of the sensitive text of its corresponding category.

[0036] To enhance the security of vector signatures, this application extracts text data feature values ​​and embeds them into the vector path of the vector signature, thus preventing information leakage without affecting the appearance of the signature. The text data to be signed is used as the associated information carrier of the vector signature. If the text data is tampered with through synonym replacement or other methods, it will directly affect the verification and recognition results. In view of the unstable features in the text that are easily modified by malicious means, the ability to identify semantic tampering is improved by establishing a logical mapping between the signature and the text semantics. The specific analysis and processing process is as follows:

[0037] First, in the text data to be signed, this application uses the BIO annotation method to mark all texts in the signed text data into two categories, namely high-sensitivity text and low-sensitivity text; for example, the fixed expressions of legal provisions and industry standard terms in contract documents are classified as low-sensitivity text because the expressions are fixed and the possibility of replacement and tampering is low; and texts containing information such as amounts and signatories that are frequently modified are classified as high-sensitivity text.

[0038] All pre-processed text data to be signed is used as input, and the TF-IDF algorithm is used to perform word frequency statistics on the word segmentation results. The word frequency of each word in all the text data to be signed is obtained as the first eigenvalue of each word, reflecting the overall word frequency characteristics of each word in all the text data to be signed. To further refine the sensitive features within different text data, the word frequency of each word in the text data to be signed is used as the second eigenvalue of each word, which is used to reflect the local word frequency characteristics of each word in the text data to be signed. For each word in a single text data to be signed, the average of the first eigenvalue and the second eigenvalue of each word is used as the word frequency statistical feature value of each word. The larger the word frequency statistical feature value, the more significant the word frequency characteristics of the corresponding word in the current text and historical texts, and the less likely the word is to undergo synonymous substitution and modification; conversely, the more likely the word is to undergo synonymous substitution and modification.

[0039] Based on the above calculation results, the mean of the word frequency statistical feature values ​​of all words in the low-sensitivity text is recorded as the first feature coefficient of the low-sensitivity text, and the mean of the word frequency statistical feature values ​​of all words in the high-sensitivity text is recorded as the second feature coefficient of the high-sensitivity text. To highlight the difference in word frequency characteristics between the two types of text and reduce the interference of synonym replacement on sensitivity feature analysis during the block segmentation process, the sum of the first feature coefficient and the second feature coefficient is calculated. The ratio of the first feature coefficient to this sum is recorded as the feature weight of the low-sensitivity text, and the ratio of the second feature coefficient to this sum is recorded as the feature weight of the high-sensitivity text.

[0040] Based on the weights determined above, a weighted adjustment is performed on the word frequency features of each word affected by synonym replacement. Specifically, the product of the word frequency statistical feature value corresponding to each word and the feature weight of the sensitive text of its corresponding category is calculated, and the product is used as the word frequency feature value of each word. The larger the word frequency feature value, the greater the impact of the synonym replacement of the word vector of the corresponding word under the hash mapping of the text data to be signed, that is, according to the comprehensive analysis of semantic features and word frequency features, the greater the possibility of synonym replacement occurring and thus affecting the hash calculation result.

[0041] Step 3: Based on the distance between the corresponding word vectors of each word and every other word in the text data to be signed and the difference in word frequency feature values, obtain the semantic feature distance and word frequency feature distance of each word respectively, and use them as the horizontal and vertical coordinates of each word to map each word to a two-dimensional coordinate system; based on the distance between any two words in the two-dimensional coordinate system and the distance between their corresponding word vectors, cluster the words in the two-dimensional coordinate system and divide all words into multiple clusters; based on the average level of the semantic feature distance and the average level of the word frequency feature distance of all words in each cluster, select some words from all the text data to be signed for block division, and then generate a vector signature.

[0042] Furthermore, considering that in the process of hash mapping the text data to be signed to form a hash value, the words that have a large difference in semantic features and word frequency features compared to the overall text data are more sensitive to synonym replacement to affect text recognition. If synonym replacement occurs, the word frequency features and semantic feature differences are combined, and the replaced words will have a larger feature change compared to the overall text. Therefore, in order to highlight the sensitivity of different words to the hash mapping changes of the text data to be signed when synonym replacement is performed, the DTW distance of the corresponding word vector between each word and each other word in the text data to be signed is calculated, and the average of all DTW distances is used as the semantic feature distance of each word; and the absolute difference in the word frequency feature value between each word and each other word in the text data to be signed is calculated, and the average of all the absolute values ​​is used as the word frequency feature distance of each word. The larger the semantic feature distance and word frequency feature distance, the greater the sensitivity of the current word to the change of semantic features in the text data to be signed, and the more likely the word is to be tampered with due to synonym replacement.

[0043] A two-dimensional coordinate system is constructed using the above-mentioned semantic feature distance and word frequency feature distance as the horizontal and vertical coordinates, respectively. All words are mapped to the constructed two-dimensional coordinate system according to their semantic feature distance and word frequency feature distance. In order to accurately reflect the semantic similarity between different words, as well as the differences in semantic features and word frequency features between different words compared to the overall text data, this embodiment calculates the similarity replacement distance of different words based on the mapping results. The specific calculation relationship is: Where, Indicates the and The similarity substitution distance between words, Indicates the and The Euclidean distance between words, Indicates the and The DTW distance between the word vectors corresponding to the words.

[0044] If the calculated similarity replacement distance is larger, it means that the semantics between different words are closer, and the semantic features and word frequency features of the text data to be signed are closer than the overall semantic features; therefore, the mapping results of all words in the two-dimensional coordinate system are used as the input of the CURE (Clustering using representative) clustering algorithm, and the similarity replacement distance between each word is used as the metric distance. The size of the shrinkage factor is set to 0.5 to obtain the clustering results of all words. In this application, the similarity replacement distance between words is used as the metric distance in the clustering division process in order to reduce the risk of attackers performing synonymous substitutions on the text to be signed, thereby tampering with the text. The specific clustering division process of the CURE algorithm is a technology well known to those skilled in the art, and the detailed process will not be repeated here.

[0045] Furthermore, in order to further highlight the sensitive features of text content replacement during the hash mapping process of the text, this embodiment conducts a comprehensive analysis of the quantity and word frequency features between different clusters based on the above clustering results. If the semantic relative features and word frequency relative features of the words in the cluster after division are more significant, the words in the cluster should be selected first when performing word segmentation.

[0046] Specifically, the priority coefficient of each cluster is calculated, and the specific calculation formula is: Where, represents the priority coefficient of the j-th cluster, represents the mean of the semantic feature distances of all words in the jth cluster, Represents the mean of the frequency feature distances of all words in the jth cluster. That is, the greater the difference in semantic features and frequency features of the corresponding words in the jth cluster, the more likely the words in that cluster are to be tampered with due to synonymous substitution, and therefore, more attention should be paid to the words in that cluster during the hash mapping process.

[0047] Hash mapping is the calculation of round functions based on data blocks, that is, the calculation result of the previous block is used as the input vector for the calculation of the next block. Therefore, in order to improve the sensitivity of synonym replacement recognition in the hash mapping process, the application determines the order of cluster selection based on the priority coefficients of all clusters, in descending order of priority coefficients. The order of selection of words within the clusters is selected in descending order according to the product of the semantic feature distance and the word frequency feature distance corresponding to each word. The selected words are sequentially placed into a preset number of blocks until the size of each block meets the preset fixed block size.

[0048] It should be noted that the purpose of taking words that have a greater impact on the sensitivity of hash mapping as the priority selected data for block segmentation is to allocate words with greater sensitivity to blocks first. Under the premise of round function calculation, when synonym replacement occurs during the hash calculation process, it will have a greater impact on the hash calculation result, thereby improving the sensitivity of tampering identification of the text data to be signed.

[0049] Hash Mapping Algorithm This embodiment uses the SHA3-512 algorithm to perform hash mapping calculations. Its fixed block size is 512 bits, and the size of the output hash value is 512 bits. First, the SHA3-512 algorithm is used to iteratively hash the words in the divided blocks to generate a 512-bit hash value, namely the hash value. The mapping process enhances the semantic tampering detection and identification capabilities by integrating the sensitive feature values ​​of the words. The hash value is then converted into a 128-bit floating-point number sequence and implicitly embedded by fine-tuning the coordinate parameters of the vector signature template path (such as the four decimal places of the coordinates of the Bezier curve control point). For example, every 4 bits of the hash value are mapped to the decimal number of the coordinate suffix (for example, the hash value "a1b2" is converted to "0.1234"). At the same time, non-visible paths (such as seal auxiliary lines) are used for redundant storage to avoid affecting the visual effect. Finally, AES-256 symmetric encryption and user public key asymmetric encryption are combined to encrypt the text data to be signed, and the vector graphics embedded with the hash value are bound to the encrypted text. The visual vector signature is generated through position anchoring and transparency control, ensuring the concealment of the hash value embedding and the integrity of the signature appearance.

[0050] See also Figure 2 , Figure 2 This is a schematic diagram of the structure of a vector signature device for preventing information leakage provided by an embodiment of the present application. In this embodiment, the terminal includes various units for executing various steps in an embodiment corresponding to a vector signature method for preventing information leakage. Figure 2 ,The vector signature device includes : a data acquisition module, a semantic word frequency ,analysis module, and a vector signature generation module.

[0051] Data collection module, used to obtain all text data to be signed;

[0052] Semantic word frequency analysis module, used to obtain the word frequency feature value of each word based on the semantic sensitivity feature and word frequency feature of each word;

[0053] The vector signature generation module is used to divide each word into multiple clusters based on the semantic features and word frequency features of each word, and select some words for block division according to the priority of each cluster and the priority of each word in each cluster, and then generate vector signatures based on the words in the block after division.

[0054] Based on the same inventive concept as the above method, an embodiment of the present application also provides a vector signature system for preventing information leakage, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-mentioned vector signature methods for preventing information leakage.

[0055] The various embodiments in this application are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

[0056] It should be noted that, unless otherwise specified and limited, terms such as "include", "comprising" or any other variations thereof are intended to cover non-exclusive inclusion, so that a circuit structure, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such article or device. In the absence of further restrictions, the phrase "including a ..." defines an element, does not exclude the presence of other identical elements in the article or device including the element. In addition, the term "and\or" used herein includes any and all combinations of one or more related listed items.

[0057] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention herein. This application is intended to cover any variations, uses, or adaptations of the present application that follow the general principles of the present application and include common knowledge or customary techniques in the art not invented herein.

[0058] It will be understood that the present application is not limited to the exact construction that has been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof.

Claims

1. A vector signature method for preventing information leakage, characterized in that: The method comprises the following steps: Get all the text data to be signed; Divide all text data to be signed into highly sensitive text and less sensitive text; obtain the word frequency statistical feature value of each word based on its word frequency in the text data to be signed and its word frequency in all text data to be signed; obtain the feature weight of the single-category sensitive text based on the ratio of the average level of the word frequency statistical feature values ​​of all words in the single-category sensitive text to the sum of the average levels of the word frequency statistical feature values ​​of all words in the two-category sensitive text; obtain the word frequency feature value of each word based on the word frequency statistical feature value of each word and the feature weight of the sensitive text of its corresponding category; Based on the distance between each word and the corresponding word vector of each other word in the text data to be signed and the difference in word frequency feature values, the semantic feature distance and word frequency feature distance of each word are obtained respectively, and each word is mapped to a two-dimensional coordinate system using them as the horizontal and vertical coordinates of each word; based on the distance between any two words in the two-dimensional coordinate system and the distance between their corresponding word vectors, the words in the two-dimensional coordinate system are clustered, and all words are divided into multiple clusters; based on the average level of the semantic feature distance and the average level of the word frequency feature distance of all words in each cluster, some words are selected from all the text data to be signed for block division, and then vector signatures are generated; The process of obtaining the semantic feature distance and word frequency feature distance of each word is as follows: the mean of the DTW distances between the corresponding word vectors of each word and each other word in the text data to be signed is used as the semantic feature distance of each word; the mean of the absolute differences in the word frequency feature values ​​between each word and each other word in the text data to be signed is used as the word frequency feature distance of each word; The specific process of dividing all words into multiple clusters is as follows: according to the distance between any two words in the two-dimensional coordinate system and the distance between their corresponding word vectors, the similarity substitution distance between the two words is obtained; the coordinates of all words in the two-dimensional coordinate system are used as the input of the clustering algorithm, and the similarity substitution distance between each word is used as the metric distance of the clustering algorithm to obtain the clustering results of all words; wherein, the calculation formula of the similarity substitution distance between the two words is: Where, Indicates the and The similarity substitution distance between words, Indicates the and The Euclidean distance between words, Indicates the and The DTW distance between the word vectors corresponding to the vocabulary.

2. A vector signature method for preventing information leakage according to claim 1, characterized in that: The process of obtaining the word frequency statistical feature value of each word is as follows: the word frequency of each word in all the text data to be signed is used as the first feature value of each word; the word frequency of each word in the text data to be signed is used as the second feature value of each word; the average of the first feature value and the second feature value of each word is used as the word frequency statistical feature value of each word.

3. A vector signature method for preventing information leakage according to claim 1, characterized in that: The process of obtaining the feature weight of the single-category sensitive text is as follows: recording the average of the word frequency statistical feature values ​​of all words in the low-sensitivity text as the first feature coefficient of the low-sensitivity text, and recording the average of the word frequency statistical feature values ​​of all words in the high-sensitivity text as the second feature coefficient of the high-sensitivity text; calculating the sum of the first feature coefficient and the second feature coefficient, recording the ratio of the first feature coefficient to the sum as the feature weight of the low-sensitivity text, and recording the ratio of the second feature coefficient to the sum as the feature weight of the high-sensitivity text.

4. A vector signature method for preventing information leakage according to claim 1, characterized in that: The word frequency feature value of each word is the product of the word frequency statistical feature value of each word and the feature weight of the sensitive text of the corresponding category.

5. A vector signature method for preventing information leakage according to claim 1, characterized in that: The specific process of selecting some words from all the text data to be signed and dividing them into blocks is as follows: Calculate the priority coefficient of each cluster: Where, represents the priority coefficient of the j-th cluster, represents the mean of the semantic feature distances of all words in the jth cluster, Represents the mean of the frequency feature distances of all words in the j-th cluster; The order of cluster selection is determined according to the priority coefficients of all clusters from large to small. The order of vocabulary selection within the cluster is selected according to the product of the semantic feature distance and the word frequency feature distance corresponding to each vocabulary from large to small. The selected vocabulary is placed in a preset number of blocks in turn until the size of each block meets the preset fixed block size.

6. A vector signature method for preventing information leakage according to claim 1, characterized in that: The specific process of generating a vector signature is as follows: performing iterative hash calculation on the words in the block after segmentation to generate a hash value; implicitly embedding the hash value into the vector signature template; using AES-256 symmetric encryption and user public key asymmetric encryption to encrypt the text data to be signed, binding the vector graphics embedded with the hash value to the encrypted text, and generating a visual vector signature through position anchoring and transparency control.

7. A vector signature device for preventing information leakage, characterized in that: Implementing a vector signature method for preventing information leakage according to any one of claims 1 to 6, the vector signature device comprises: Data collection module, used to obtain all text data to be signed; Semantic word frequency analysis module, used to obtain the word frequency feature value of each word based on the semantic sensitivity feature and word frequency feature of each word; The vector signature generation module is used to divide each word into multiple clusters based on the semantic features and word frequency features of each word, and select some words for block division according to the priority of each cluster and the priority of each word in each cluster, and then generate vector signatures based on the words in the block after division.

8. A vector signature system for preventing information leakage, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the vector signature method for preventing information leakage according to any one of claims 1 to 6 is implemented.

Images