A blockchain-based digital asset security assessment method

Abstract

The application relates to the technical field of information security management, in particular to a digital asset security evaluation method based on a block chain, which comprises the following steps: collecting user historical data and dividing the user historical data into local information sections according to time, obtaining action word section word frequency and label coefficients through processing and analysis, and then determining user labels; integrating user information of the same label to form an account set, generating individual data output curves based on user data in the set, refining comprehensive curve features of the target set, and finally calculating output speed through real-time monitoring of user output data, combining the comprehensive feature curve rate point of the corresponding label, calculating the rate offset risk value to identify abnormal users. The application improves the accuracy of label determination, the rationality of account integration, the comprehensiveness of feature analysis and the efficiency of abnormal identification, and effectively guarantees the security of block chain digital assets.

Description

Technical Field

[0001] This invention relates to the field of information security management technology, and in particular to a blockchain-based digital asset security assessment method. Background Technology

[0002] With the rapid development of information technology, blockchain technology has been widely used in many fields due to its decentralized, immutable, and highly secure characteristics, including the field of digital assets.

[0003] Existing technology CN119417612A discloses a method and system for assessing the security risks of digital assets. This includes obtaining the unique identifier and ownership record of the digital asset, identifying key points for risk assessment (including encrypted storage, transaction history, and circulation mechanism), identifying identity authentication, access control, and data encryption security factors for each key point, and using probabilistic statistical methods to verify the security compliance of the digital asset's privacy protection and compliance requirements. Through Monte Carlo simulation, multiple combinations of cross-chain compatibility and value assessment factors for digital assets are generated to simulate security events and their consequences under different scenarios, calculating the probability of occurrence and impact of loss for each scenario. Based on the optimized risk assessment model, the overall security risk of the digital asset is assessed, the highest-risk critical link is identified, and its potential impact of loss is evaluated.

[0004] However, in the current blockchain environment, digital asset security faces the challenge of a surge in data diversity and complexity. The massive amount of user information lacks systematic classification and in-depth analysis methods, making it difficult to uncover potential risks. Traditional user tagging methods are singular, relying only on a few features or single-dimensional information, which cannot fully reflect the user's true attributes. This leads to biases in the account sets formed by integrating user information with the same tags. At the same time, in the security assessment stage, existing technologies focus more on the surface features of data, ignoring the dynamic changes in data output and user correlations, making it impossible to identify abnormal users in a timely manner. Summary of the Invention

[0005] The purpose of this invention is to address the problems in the background art by proposing a blockchain-based digital asset security assessment method.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A blockchain-based digital asset security assessment method, which specifically includes the following steps:

[0008] Step 1: Identify the data assets existing in the target blockchain. Data assets refer to user information. Collect historical data generated in the user information and divide it into local information segments according to the time of data generation. Then, process and analyze the information in the local information segments to obtain action word segments and word frequencies. Calculate the word frequencies of the action word segments to obtain the tag coefficients of the action word segments, and determine the user's user tags based on the tag coefficients.

[0009] Step 2: Identify all user information and corresponding user tags in the target blockchain, and integrate user information with the same user tags to obtain an account set;

[0010] Step 3: Sequentially set the account set as the target set. Based on all user information in the target set, obtain the data production information of each user and analyze the data production information to determine the user's data output curve. Then, determine the comprehensive characteristic curve of the target set based on the data output curve.

[0011] Step 4: Monitor user information in the target blockchain in real time to obtain user output data, calculate the data output rate based on the output data, identify the user tags of the corresponding users, determine the rate point according to the comprehensive feature curve corresponding to the user tags, combine the data output rate with the rate point to calculate the rate deviation risk value, and determine the abnormal users of the account based on the rate deviation risk value.

[0012] As a further aspect of the present invention, the method for obtaining action segments and word frequencies includes:

[0013] S1: Identify user information existing in the target blockchain, mark the target users with the user information in sequence, extract the data generated by the target users within a fixed period, and mark this data as the specified analysis data;

[0014] S2: Use a text conversion algorithm to convert all information in the specified analysis data into text form;

[0015] The system acquires the converted specified analysis data, identifies the generation time of the specified analysis data, sets a cycle period, segments the specified analysis data according to the generation time of the specified analysis data according to the cycle period, and marks the specified analysis data within the cycle period as a local information segment. Furthermore, one cycle period corresponds to one local information segment.

[0016] S3: Obtain the local information segment and split the continuous sentences in the local information segment into independent words. At the same time, mark the part of speech of each word, which includes content words and function words. Extract all words with the part of speech of content words in the local information segment and mark them as action segments. First, count the total number of action segments AL. Then, integrate all action segments and count the number of times each action segment appears Ci, where i represents different action segments. Then, use the formula fi=Ci÷AL to get the word frequency fi of action segment i.

[0017] Using the method described above, identify the action word segments present in each local information segment and calculate the word frequency of each action word segment in the local information segment.

[0018] As a further aspect of the present invention, content words refer to words that have actual meaning and can function as sentence components on their own, while function words refer to words that have no actual meaning and only serve as grammatical connectors. At the same time, the total number AL of action words in each local information segment is greater than or equal to the minimum sample value. If the total number AL of action words in a local information segment is less than the sample value, the corresponding local information segment is deleted.

[0019] As a further aspect of the present invention, the method for determining user tags includes:

[0020] The action segments are sequentially labeled as target words. Local information segments containing target words are identified, and the word frequency of the target words in the corresponding local information segments is determined. j represents different local information segments, and j∈[1,n], indicating that there are a total of n local information segments containing the target word;

[0021] Using formula The discrete values ​​U of the target word are obtained, where, for The mean;

[0022] Based on the discrete value U and the mean Then use the formula The tag coefficient B of the target word is obtained, where m represents the number of local information segments in which the target word appears, and Ad represents the total number of local information segments;

[0023] Obtain all action words and corresponding tag coefficients for the target user. Then compare the tag coefficient B with the threshold X1. If B ≥ X1, set the corresponding action word as the user tag for this target user. Otherwise, if B < X1, do not set the corresponding action word.

[0024] As a further aspect of the present invention, the method for determining the data output curve includes:

[0025] Based on the users existing in the target set, and according to the continuous online time of each user, a continuous online time is marked as an independent online time. At this time, a user corresponds to multiple independent online times.

[0026] Set a unit time, arbitrarily select an independent online time, and start from the beginning of this independent online time to count the amount of data generated by the corresponding user in each unit time within this independent online time.

[0027] Using time as the x-axis and data volume as the y-axis, a planar coordinate system is set up. Then, the data volume generated in each unit of time during the independent online time is marked on the planar coordinate system to obtain several unit points. Then, using linear fitting technology, multiple unit points are fitted into a curve, and this curve is marked as the data output curve. Here, one independent online time corresponds to one data output curve, and the data volume corresponding to the endpoint of the data output curve is the total data volume generated by the corresponding user in the independent online time.

[0028] As a further aspect of the present invention, the method for determining the comprehensive characteristic curve includes:

[0029] Using image tools, we simulated the data output curve to determine the functional expression HD of the data output curve;

[0030] The rate change function is obtained by performing a first derivative operation on the function expression HD. At the same time, the planar coordinate system is reset, and the rate of change is adjusted according to the function. Plot the curve in the newly set planar coordinate system to obtain the rate curve;

[0031] Calculate the rate change function for all independent online times in the target set, and plot the corresponding rate curves in the plane coordinate system according to the rate change function. At this time, there are several rate curves in the plane coordinate system. Then, use the recurrent neural network algorithm to extract common features among the several rate curves and mark them as the comprehensive feature curve of the target set.

[0032] As a further aspect of the present invention, the duration of independent online time is greater than or equal to the minimum time threshold. If there is an independent online time that is less than the time threshold, the corresponding independent online time will not be included in the data output feature processing.

[0033] As a further aspect of the present invention, the method for obtaining the rate offset risk value includes:

[0034] Real-time data monitoring of user information in the target blockchain is performed to obtain user output data. Then, the user tags of the corresponding users are identified, and the corresponding account set and comprehensive feature curve are selected based on the user tags.

[0035] Get the unit time again, and according to the unit time, start from the user's initial online time, count the output data in each unit time, divide the output data in each unit time by the unit time to get the real-time data output speed VSp, where p represents different unit time, and p∈[1,k], and k represents the total number of unit time generated by the user at the current time;

[0036] Next, obtain the corresponding comprehensive characteristic curve, and determine the corresponding rate point VBp on the comprehensive characteristic curve according to the node corresponding to each unit time, using the formula... The real-time rate offset risk value Y is obtained.

[0037] As a further aspect of the present invention, the method for determining users with abnormal accounts includes:

[0038] The real-time rate offset risk value Y is compared with the risk threshold Z1 in this account set. If Y≤Z1, a safety signal is generated; otherwise, if Y>Z1, an abnormal risk signal is generated.

[0039] When the system detects an abnormal risk signal, it retrieves the user corresponding to the abnormal risk signal, marks this user as an account abnormal user, restricts the account permissions of the account abnormal user, and generates a security abnormal signal notification to the account abnormal user.

[0040] Compared with existing technologies, the advantages of this invention are:

[0041] This invention collects historical user data and divides it into local information segments according to time. After processing and analysis, it obtains the frequency of action words and tag coefficients to determine user tags. Then, it integrates user information with the same tags to form an account set. Based on the user data production information within the set, it generates individual data output curves and then extracts the comprehensive feature curve of the target set. Finally, it calculates the output speed by monitoring user output data in real time and combines the rate point of the comprehensive feature curve of the corresponding tag to calculate the rate deviation risk value to identify abnormal users. This solution improves the accuracy of user tags by dynamically dividing data segments and multi-dimensional word frequency analysis, achieves scientific integration of account sets based on accurate tags, captures data output patterns by modeling individual and comprehensive feature curves, and achieves rapid identification of abnormal users by using real-time rate monitoring and deviation calculation. This improves the accuracy of tag determination, the rationality of account integration, the comprehensiveness of feature analysis, and the efficiency of anomaly identification, effectively ensuring the security of blockchain digital assets. Attached Figure Description

[0042] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0043] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0044] Reference Figure 1 A blockchain-based digital asset security assessment method, which specifically includes the following steps:

[0045] Step 1: Select the target blockchain and identify the digital assets in the target blockchain. Here, digital assets refer to assets that exist in binary digital form and have ownership or value. In this embodiment, the digital assets are set as user information.

[0046] The process involves acquiring and analyzing user information to determine user tags. Specific methods for determining user tags include:

[0047] S1: Identify user information existing in the target blockchain, arbitrarily select a user information, take this user information as an example, mark this user information as the target user, extract the data generated by the target user within a fixed period, and mark this data as the specified analysis data;

[0048] The specific time range of the fixed period is set by those skilled in the art based on big data experience. In this embodiment, the fixed period is set to six months.

[0049] S2: Use a text conversion algorithm to convert all information in the specified analysis data into text form;

[0050] In this embodiment, the text conversion algorithm uses natural language generation technology, which is an existing technology and will not be described in detail here.

[0051] The system acquires the converted specified analysis data, identifies the generation time of the specified analysis data, sets a cycle period, segments the specified analysis data according to the generation time of the specified analysis data according to the cycle period, and marks the specified analysis data within the cycle period as a local information segment. Furthermore, one cycle period corresponds to one local information segment.

[0052] The specific time range of the cycle is set by those skilled in the art based on big data experience. In this embodiment, the cycle is set to one week. Furthermore, the time range of the cycle is shorter than the time range of the fixed period.

[0053] S3: Obtain a local information segment and break down the continuous sentences in the local information segment into independent words. At the same time, mark the part of speech of each word. The part of speech includes content words and function words. Furthermore, content words refer to words that have actual meaning and can stand alone as sentence components, while function words refer to words that have no actual meaning and only serve as grammatical connectors.

[0054] Extract all words with the part of speech of content words from the local information segment and mark them as action segments. First, count the total number of action segments AL. Then, integrate all action segments and count the number of times each action segment appears Ci, where i represents different action segments. Then, use the formula fi=Ci÷AL to get the word frequency fi of action segment i.

[0055] Following the above method, identify the action word segments present in each local information segment and calculate the word frequency of each action word segment in the local information segment;

[0056] It should be further explained that the total number of action words AL in each local information segment is greater than or equal to the minimum sample value. The specific value of the minimum sample value is set by those skilled in the art based on big data experience. If the total number of action words AL in a local information segment is less than the sample value, the corresponding local information segment will be deleted.

[0057] S4: Randomly select an action word segment, take this action word segment as an example, and mark this action word segment as the target word. Then, identify the local information segments containing the target word and identify the word frequency of the target word in the corresponding local information segments. , where j represents different local information segments, and j∈[1,n], indicating that there are a total of n local information segments containing the target word;

[0058] Using formula The discrete values ​​U of the target word are obtained, where, for The mean;

[0059] Then, based on the discrete value U and the mean... Then use the formula The tag coefficient B of the target word is obtained, where m represents the number of local information segments in which the target word appears, and Ad represents the total number of local information segments. and These are the proportionality coefficients, and The specific values ​​were obtained by those skilled in the art through big data calculations. In this embodiment, 0 < <1, 0< <1, and ;

[0060] Furthermore, the larger the tag coefficient U of the target word, the higher the representativeness of the target word to the target user; conversely, the smaller the tag coefficient U of the target word, the lower the representativeness of the target word to the target user.

[0061] All action segments are sequentially labeled as target words, and the label coefficient B is calculated using the method described above;

[0062] S5: Obtain all action segments and corresponding tag coefficients of the target user, and then compare the tag coefficient B with the threshold X1. If B≥X1, the corresponding action segment is set as the user tag of this target user. Otherwise, if B<X1, the corresponding action segment is not set. The specific value of the threshold X1 is obtained by those skilled in the art based on big data calculation.

[0063] Step 2: Identify all user information and corresponding user tags in the target blockchain, and integrate user information with the same user tag to obtain an account set, where one account set corresponds to one user tag;

[0064] Step 3: Select any user tag. Taking this user tag as an example, mark this user tag as the target tag. At the same time, obtain the set of accounts corresponding to this target tag and mark them as the target set.

[0065] Based on all user information in the target set, data production information is obtained from each user's information and analyzed to determine the comprehensive characteristic curve of the target set. Specific methods for determining the comprehensive characteristic curve include:

[0066] SS1: Based on the users existing in the target set, and according to the continuous online time of each user, a continuous online time is marked as an independent online time. At this time, a user corresponds to multiple independent online times.

[0067] SS2: Set unit time. Choose any independent online time. Starting from the beginning of this independent online time, count the amount of data generated by the corresponding user in each unit time within this independent online time. The specific time of the unit time shall be set by those skilled in the art based on big data experience.

[0068] Using time as the x-axis and data volume as the y-axis, a planar coordinate system is set up. Then, the data volume generated in each unit of time during the independent online time is marked on the planar coordinate system to obtain several unit points. Then, using linear fitting technology, multiple unit points are fitted into a curve, and this curve is marked as the data output curve. Furthermore, one independent online time corresponds to one data output curve, and the data volume corresponding to the endpoint of the data output curve is the total data volume generated by the corresponding user in the independent online time.

[0069] SS3: Using image tools, perform simulation calculations on the data output curve to determine the functional expression HD of the data output curve. In this embodiment, the image tool used is Matlab. The specific operation method is existing technology and will not be described in detail here.

[0070] The rate change function is obtained by performing a first derivative operation on the function expression HD. At the same time, the planar coordinate system is reset, and the rate of change is adjusted according to the function. Plot the curve in the newly set planar coordinate system to obtain the rate curve;

[0071] Following the above method, calculate the rate change function of all independent online times in the target set, and plot the corresponding rate curve in the plane coordinate system according to the rate change function. At this time, there are several rate curves in the plane coordinate system. Then, use the recurrent neural network algorithm to extract common features among the several rate curves and mark them as the comprehensive feature curve of the target set. The recurrent neural network algorithm is an existing technology, and the specific processing process will not be described here.

[0072] It should be further noted that the duration of independent online time is greater than or equal to the minimum time threshold. If there is an independent online time that is less than the time threshold, the corresponding independent online time will not be included in the data output feature processing.

[0073] All account sets are sequentially marked as target sets, and processed according to the above method to determine the comprehensive characteristic curve of each account set;

[0074] Step 4: Real-time data monitoring of user information in the target blockchain to obtain user output data, then identify the corresponding user tags, and select the corresponding account set and comprehensive feature curve based on the user tags.

[0075] Get the unit time again, and according to the unit time, start from the user's initial online time, count the output data in each unit time, divide the output data in each unit time by the unit time to get the real-time data output speed VSp, where p represents different unit time, and p∈[1,k], and k represents the total number of unit time generated by the user at the current time;

[0076] Next, obtain the corresponding comprehensive characteristic curve, and determine the corresponding rate point VBp on the comprehensive characteristic curve according to the node corresponding to each unit time, using the formula... Obtain the real-time rate offset risk value Y;

[0077] The real-time rate offset risk value Y is compared with the risk threshold Z1 in this account set. If Y≤Z1, a safety signal is generated; otherwise, if Y>Z1, an abnormal risk signal is generated. The specific value of the risk threshold Z1 is obtained by those skilled in the art based on big data calculations.

[0078] When the system detects an abnormal risk signal, it retrieves the user corresponding to the abnormal risk signal, marks this user as an account abnormal user, restricts the account permissions of the account abnormal user, and generates a security abnormal signal notification to the account abnormal user.

[0079] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A blockchain-based digital asset security assessment method, characterized in that, The method specifically includes the following steps: Step 1: Identify the data assets existing in the target blockchain. Data assets refer to user information. Collect historical data generated in the user information and divide it into local information segments according to the time of data generation. Then, process and analyze the information in the local information segments to obtain action word segments and word frequencies. Calculate the word frequencies of the action word segments to obtain the tag coefficients of the action word segments, and determine the user's user tags based on the tag coefficients. Step 2: Identify all user information and corresponding user tags in the target blockchain, and integrate user information with the same user tags to obtain an account set; Step 3: Sequentially set the account set as the target set. Based on all user information in the target set, obtain the data production information of each user and analyze the data production information to determine the user's data output curve. Then, determine the comprehensive characteristic curve of the target set based on the data output curve. Methods for determining data output curves and comprehensive characteristic curves include: Based on the users existing in the target set, and according to the continuous online time of each user, a continuous online time is marked as an independent online time. At this time, a user corresponds to multiple independent online times. Set a unit time, arbitrarily select an independent online time, and start from the beginning of this independent online time to count the amount of data generated by the corresponding user in each unit time within this independent online time. Using time as the x-axis and data volume as the y-axis, a planar coordinate system is set up. Then, the data volume generated in each unit of time during the independent online time is marked on the planar coordinate system to obtain several unit points. Then, using linear fitting technology, multiple unit points are fitted into a curve, and this curve is marked as the data output curve. Here, one independent online time corresponds to one data output curve, and the data volume corresponding to the endpoint of the data output curve is the total data volume generated by the corresponding user in the independent online time. Using image tools, we simulated the data output curve to determine the functional expression HD of the data output curve; The rate change function is obtained by performing a first derivative operation on the function expression HD. At the same time, the planar coordinate system is reset, and the rate of change is adjusted according to the function. Plot the curve in the newly set planar coordinate system to obtain the rate curve; Calculate the rate change function for all independent online times in the target set, and plot the corresponding rate curves in the plane coordinate system according to the rate change function. At this time, there are several rate curves in the plane coordinate system. Then, use the recurrent neural network algorithm to extract common features among the several rate curves and mark them as the comprehensive feature curve of the target set. Step 4: Monitor user information in the target blockchain in real time to obtain user output data, calculate the data output rate based on the output data, identify the user tags of the corresponding users, determine the rate point according to the comprehensive feature curve corresponding to the user tags, combine the data output rate with the rate point to calculate the rate deviation risk value, and determine the abnormal users of the account based on the rate deviation risk value.

2. The blockchain-based digital asset security assessment method according to claim 1, characterized in that, Methods for obtaining action phrases and word frequencies include: S1: Identify user information existing in the target blockchain, mark the target users with the user information in sequence, extract the data generated by the target users within a fixed period, and mark this data as the specified analysis data; S2: Use a text conversion algorithm to convert all information in the specified analysis data into text form; The system acquires the converted specified analysis data, identifies the generation time of the specified analysis data, sets a cycle period, segments the specified analysis data according to the generation time of the specified analysis data according to the cycle period, and marks the specified analysis data within the cycle period as a local information segment. Furthermore, one cycle period corresponds to one local information segment. S3: Obtain the local information segment and split the continuous sentences in the local information segment into independent words. At the same time, mark the part of speech of each word, which includes content words and function words. Extract all words with the part of speech of content words in the local information segment and mark them as action segments. First, count the total number of action segments AL. Then, integrate all action segments and count the number of times each action segment appears Ci, where i represents different action segments. Then, use the formula fi=Ci÷AL to get the word frequency fi of action segment i. Using the method described above, identify the action word segments present in each local information segment and calculate the word frequency of each action word segment in the local information segment.

3. The blockchain-based digital asset security assessment method according to claim 2, characterized in that, Content words refer to words that have actual meaning and can function as sentence components on their own, while function words refer to words that have no actual meaning and only serve as grammatical connectors. At the same time, the total number of action words (AL) in each local information segment is greater than or equal to the minimum sample value. If the total number of action words (AL) in a local information segment is less than the sample value, the corresponding local information segment will be deleted.

4. The blockchain-based digital asset security assessment method according to claim 2, characterized in that, Methods for determining user tags include: The action segments are sequentially labeled as target words. Local information segments containing target words are identified, and the word frequency of the target words in the corresponding local information segments is determined. j represents different local information segments, and j∈[1,n], indicating that there are a total of n local information segments containing the target word; Using formula The discrete values ​​U of the target word are obtained, where, for The mean; Based on the discrete value U and the mean Then use the formula The tag coefficient B of the target word is obtained, where m represents the number of local information segments in which the target word appears, and Ad represents the total number of local information segments. and These are the proportionality coefficients, 0 < <1, 0< <1, and ; Obtain all action words and corresponding tag coefficients for the target user. Then compare the tag coefficient B with the threshold X1. If B ≥ X1, set the corresponding action word as the user tag for this target user. Otherwise, if B < X1, do not set the corresponding action word.

5. The blockchain-based digital asset security assessment method according to claim 1, characterized in that, The duration of independent online time is greater than or equal to the minimum time threshold. If the duration of independent online time is less than the time threshold, the corresponding independent online time will not be included in the data output feature processing.

6. The blockchain-based digital asset security assessment method according to claim 1, characterized in that, Methods for obtaining rate offset risk values ​​include: Real-time data monitoring of user information in the target blockchain is performed to obtain user output data. Then, the user tags of the corresponding users are identified, and the corresponding account set and comprehensive feature curve are selected based on the user tags. Get the unit time again, and according to the unit time, start from the user's initial online time, count the output data in each unit time, divide the output data in each unit time by the unit time to get the real-time data output speed VSp, where p represents different unit time, and p∈[1,k], and k represents the total number of unit time generated by the user at the current time; Next, obtain the corresponding comprehensive characteristic curve, and determine the corresponding rate point VBp on the comprehensive characteristic curve according to the node corresponding to each unit time, using the formula... The real-time rate offset risk value Y is obtained.

7. The blockchain-based digital asset security assessment method according to claim 6, characterized in that, Methods for identifying users with abnormal accounts include: The real-time rate offset risk value Y is compared with the risk threshold Z1 in this account set. If Y≤Z1, a safety signal is generated; otherwise, if Y>Z1, an abnormal risk signal is generated. When the system detects an abnormal risk signal, it retrieves the user corresponding to the abnormal risk signal, marks this user as an account abnormal user, restricts the account permissions of the account abnormal user, and generates a security abnormal signal notification to the account abnormal user.

Images