Data automation resource transaction management method and system

By constructing a transaction detection model based on historical transaction records, and encrypting and detecting anomalies in real-time transaction records, the problems of low accuracy in anomaly identification and privacy leakage in resource trading platforms are solved, achieving efficient anomaly detection and privacy protection.

CN119027239AInactive Publication Date: 2024-11-26HUBEI OIN NETWORK TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411250147.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-06
Publication Date
2024-11-26
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Existing technologies for identifying transaction anomalies on resource trading platforms have low accuracy and pose a risk of user privacy leaks.

Method used

By constructing a transaction detection model based on historical resource transaction records, real-time resource transaction records are preprocessed and encrypted, and anomaly detection is performed using a cloud server. A single-layer decision tree and encryption algorithm are used to protect user privacy.

Benefits of technology

It improves the accuracy and scope of anomaly detection while ensuring user privacy and preventing privacy leaks.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119027239A_ABST
    Figure CN119027239A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of transaction management, and discloses a data automation resource transaction management method and system, and the method comprises the steps: obtaining a real-time resource transaction record; preprocessing and encrypting the real-time resource transaction record to obtain a first target data set; uploading the first target data set to a cloud server, and performing anomaly detection on the first target data set by a transaction detection model deployed on the cloud server to obtain a detection result; wherein the transaction detection model is constructed based on historical resource transaction records; and obtaining a detection result of the cloud server, generating a detection report according to the detection result, and visually displaying the detection report. According to the method, the data privacy is protected, and meanwhile, the accuracy and the high efficiency of anomaly detection can be ensured.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of transaction management, and in particular relates to a data automation resource transaction management method and system. Background Art

[0002] Online resource trading refers to the provision of services for resource trading through online trading platforms, which enables buyers and sellers of resources to complete transactions in a transparent and fair environment. Resource trading types mainly include natural resources such as mining resources and energy resources, as well as data resources such as big data and information services.

[0003] The trading platform can provide supply and demand information of resources and manage transaction information during the transaction process. For example, supply and demand information can include resource type, quantity, quality, price, etc.

[0004] The construction and management of resource trading platforms play an important role in promoting the rational use of resources, protecting the ecological environment, and promoting economic development. With the advancement of technology and the diversification of market demand, resource trading platforms are also constantly innovating and upgrading.

[0005] In the actual application process, the resource trading platform will be subject to potential attack threats, which will cause economic losses to the resource trading platform. Therefore, the resource trading platform needs to implement regulatory measures such as auditing and anomaly detection during the transaction process, which is particularly important for preventing attack threats.

[0006] In the prior art, transaction anomalies are identified by analyzing the behavior of suspicious users, for example, using algorithms such as support vector machines and K-means clustering. However, these methods can only identify a small number of anomalies with low accuracy. In addition, it is difficult to ensure user privacy during the identification process, and there is a risk of user privacy leakage. Summary of the invention

[0007] The purpose of the present invention is to provide a data automation resource transaction management method and system to solve the problems in the prior art of low accuracy in identifying transaction anomalies and easy leakage of user privacy.

[0008] In order to achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a data automation resource transaction management method, the method comprising: Obtain real-time resource transaction records; Preprocessing and encrypting the real-time resource transaction records to obtain a first target data set; The first target data set is uploaded to the cloud server, and a transaction detection model deployed on the cloud server performs anomaly detection on the first target data set to obtain a detection result; wherein the transaction detection model is constructed based on historical resource transaction records; Obtain the test results of the cloud server, generate a test report based on the test results, and visualize the test report.

[0009] Preferably, the method further comprises: constructing a transaction detection model, comprising: Acquire historical resource transaction records, pre-process the historical resource transaction records, and obtain a second target data set; encrypting the second target data set to obtain an encrypted second target data set; The encrypted second target data set is sent to a cloud server, and the cloud server constructs a transaction detection model according to the encrypted second target data set.

[0010] Preferably, the historical resource transaction records are preprocessed to obtain a second target data set, including: Extract features from historical resource transaction records to obtain a second feature vector; Record the label of the historical resource transaction record to obtain the characteristic label, wherein the historical resource transaction record contains the label of the abnormal transaction; The second eigenvector is combined with the feature label to obtain a transaction record data matrix; A preset number of columns are extracted from the transaction record data matrix, feature thresholds are calculated based on all elements in each extracted column, a threshold vector is constructed based on the feature thresholds corresponding to all extracted columns, and the threshold vector, the second feature vector and the feature label are used as the second target data set.

[0011] Preferably, encrypting the second target data set to obtain the encrypted second target data set includes: The second feature vector and the threshold vector are encrypted respectively based on the public key of the encryption algorithm to obtain a feature ciphertext vector corresponding to the second feature vector and a ciphertext threshold vector corresponding to the threshold vector; Randomly generate a first reversible matrix, a second reversible matrix, a first triangular matrix, and a second triangular matrix; Generate a first diagonal matrix based on the ciphertext threshold vector; Generate a first threshold matrix based on the first reversible matrix, the first triangular matrix and the first diagonal matrix, generate a second threshold matrix based on the second reversible matrix, the second triangular matrix and the first diagonal matrix, and use the first threshold matrix and the second threshold matrix as threshold encrypted data sets; Generate a second diagonal matrix based on the characteristic ciphertext vector; A second characteristic matrix is ​​generated based on the second diagonal matrix and the first reversible matrix, and the second characteristic matrix is ​​used as a characteristic encrypted data set; and the threshold encrypted data set and the characteristic encrypted data set are used as the encrypted second target data set.

[0012] Preferably, the cloud server constructs a transaction detection model according to the encrypted second target data set, including: The cloud server calculates the data set to be decrypted based on the feature encrypted data set and the threshold encrypted data set; The cloud server decrypts the data set to be decrypted based on the private key to obtain a decrypted data set; Compare each element in the decrypted data set with a preset value to obtain a comparison result of each element of the decrypted data set; Based on the single-layer decision tree and feature labels, the comparison results of each element of the decrypted data set are divided to obtain classification labels; Determine a set of candidate base classifiers of a single-layer decision tree based on the classification labels, the comparison results of each element of the decrypted data set, and the threshold encrypted data set; Iteratively train the candidate base classifier set of the single-layer decision tree to obtain a number of base classifiers and coefficients corresponding to each base classifier; A transaction detection model is constructed based on several base classifiers and coefficients corresponding to each base classifier.

[0013] Preferably, the transaction detection model deployed on the cloud server performs anomaly detection on the first target data set to obtain a detection result, including: Generate a classification label for each element in the first target data set based on the transaction detection model; A detection result is determined based on the classification label of each element in the first target data set and the coefficient corresponding to each base classifier.

[0014] Preferably, the real-time resource transaction record is preprocessed and encrypted to obtain a first target data set, including: Extracting features from real-time resource transaction records to obtain a first feature vector; The first feature vector is encrypted to obtain a first target data set.

[0015] Preferably, encrypting the first feature vector to obtain the first target data set includes: Encrypting the first feature vector based on a public key of an encryption algorithm to obtain a first feature ciphertext vector; Generate a third diagonal matrix based on the first characteristic ciphertext vector; A first characteristic matrix is ​​generated based on the second reversible matrix and the third diagonal matrix, and the first characteristic matrix is ​​used as a first target data set.

[0016] In a second aspect, the present invention provides a data automation resource transaction management system, the system is used to implement the above-mentioned data automation resource transaction management method, the system comprises: Data acquisition module, used to obtain real-time resource transaction records; A data processing module, used for preprocessing and encrypting the real-time resource transaction records to obtain a first target data set; A data detection module is used to upload the first target data set to the cloud server, and the transaction detection model deployed on the cloud server performs anomaly detection on the first target data set to obtain a detection result; wherein the transaction detection model is constructed based on historical resource transaction records; The detection display module is used to obtain the detection results of the cloud server, generate a detection report based on the detection results, and visually display the detection report.

[0017] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned data automated resource transaction management method.

[0018] Beneficial effects: The present invention constructs a transaction detection model through historical resource transaction records, which contain a variety of sample data; and then uses the transaction detection model to perform anomaly detection on real-time resource transaction records, which can improve the accuracy and range of anomaly detection; and in the detection process, the real-time resource transaction records are encrypted, which can effectively ensure that the user's privacy will not be leaked. Therefore, the present invention can protect data privacy while ensuring the accuracy and efficiency of anomaly detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings: Figure 1 It is a flow chart of a data automation resource transaction management method provided by an embodiment of the present invention; Figure 2 It is a block diagram of a data automation resource transaction management system provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the present invention will be briefly introduced below in combination with the drawings and the description of the embodiments or the prior art. Obviously, the following description of the structure of the drawings is only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention.

[0021] Embodiment 1 Figure 1 It is a flow chart of a data automation resource transaction management method provided by an embodiment of the present invention. The application scenarios of the data automation resource transaction management method of this embodiment include: a transaction terminal, a transaction server and a cloud server. The transaction terminal and the transaction server are communicated and connected. The transaction helps both parties to complete the transaction process. It can process transaction requests from buyers or sellers, and forward these requests to corresponding transaction counterparties, as well as ensure the accurate transmission of transaction information, including price, quantity, terms, etc., to ensure that both parties to the transaction can obtain relevant information in a timely manner; wherein, the transaction terminal can be a PC, a smart phone, a platform and other smart devices, and the transaction server is accessed through an application running on the transaction terminal.

[0022] The cloud server is connected to the transaction server for communication. The cloud server is mainly used to monitor the transaction tasks generated on the transaction server and detect whether there are any abnormalities in the transaction tasks.

[0023] Based on the above application scenario, this embodiment provides a data automation resource transaction management method, which is applied on a transaction server. Figure 1 As shown, the method includes: Step S10: Acquire real-time resource transaction records.

[0024] In this embodiment, the transaction server can record the transaction records generated on each transaction terminal and store them locally on the transaction server, so that the relevant resource transaction records can be directly obtained locally on the transaction server.

[0025] Step S20: pre-processing and encrypting the real-time resource transaction records to obtain a first target data set.

[0026] In order to ensure that personal privacy data in the real-time resource transaction record will not be leaked, specifically, the real-time resource transaction record is preprocessed and encrypted to obtain a first target data set, including: Step S201: extracting features from real-time resource transaction records to obtain a first feature vector; Step S202: encrypt the first feature vector to obtain a first target data set.

[0027] In this embodiment, the transaction features of the real-time resource transaction record are extracted to obtain a first feature vector, and an element in the first feature vector is used as a transaction feature; then the first feature vector is encrypted to obtain a first target data set, and the transaction server then sends the first target data set to the cloud server for anomaly detection; therefore, before the real-time resource transaction record is sent to the cloud server, the encryption operation can effectively improve the protection of user personal privacy and avoid personal privacy leakage.

[0028] Step S30: Upload the first target data set to the cloud server, and use the transaction detection model deployed on the cloud server to perform anomaly detection on the first target data set to obtain a detection result; wherein the transaction detection model is constructed based on historical resource transaction records.

[0029] In this embodiment, the steps of constructing a transaction detection model on a cloud server are: Step a10: Acquire historical resource transaction records, pre-process the historical resource transaction records, and obtain a second target data set.

[0030] On the trading server, these historical resource transaction records are data after experts' anomaly analysis. These transaction records have corresponding labels, which are classification labels of transaction records, used to characterize whether the transaction is abnormal and the type of anomaly.

[0031] The historical resource transaction records also contain the user's personal privacy information, so the historical resource transaction records also need to be protected for privacy. Specifically, the historical resource transaction records are preprocessed to obtain a second target data set, including: Step a101: extract features from historical resource transaction records to obtain a second feature vector.

[0032] There are n historical resource transaction records stored in the transaction server. After feature extraction is performed on each historical resource transaction record and then normalization and other processing are performed on the extracted features, a second feature vector can be obtained. Each element in the second feature vector corresponds to a feature data in the historical resource transaction record.

[0033] Step a102: Record the labels of historical resource transaction records to obtain characteristic labels, wherein the historical resource transaction records contain labels of abnormal transactions.

[0034] In this embodiment, each historical resource transaction record has a label, and the label is used to reflect whether an abnormal transaction occurs in the historical resource transaction data and the type of abnormality.

[0035] Step a103: Merge the second feature vector and the feature label to obtain a transaction record data matrix.

[0036] In this embodiment, assuming that m features can be extracted from a piece of historical resource transaction data, the transaction record data matrix has n rows and m+1 columns.

[0037] Step a104: extract a preset number of columns from the transaction record data matrix, calculate the feature threshold based on all elements in each extracted column, construct a threshold vector based on the feature thresholds corresponding to all extracted columns, and use the threshold vector, the second feature vector and the feature label as the second target data set.

[0038] In this embodiment, the non-label columns of the transaction record data matrix are taken as the columns to be extracted. At this time, the preset number is m columns. For all the data in these extracted columns, the average value of all the data in each column is calculated, and then rounded (rounded down or rounded up). The rounded average value is used as the feature threshold. At this time, m feature thresholds can be obtained. Then, these m feature thresholds are used to construct a row vector to obtain a threshold vector. The m second feature vectors, a feature label and a threshold vector corresponding to each historical resource transaction record are used as a second target data set.

[0039] Step a20: encrypting the second target data set to obtain an encrypted second target data set.

[0040] In this embodiment, the steps of encrypting the second target data set are: 1) Encrypting the second feature vector and the threshold vector based on the public key of the encryption algorithm to obtain a feature ciphertext vector corresponding to the second feature vector and a ciphertext threshold vector corresponding to the threshold vector; In this embodiment, the encryption algorithm adopts the Paillier encryption algorithm. The transaction server has a public key of the Paillier encryption algorithm, and the cloud server has a private key of the Paillier encryption algorithm. The Paillier encryption algorithm generates a key in the following manner: Choose two large prime numbers p and q , calculate the modulus N , N = p × q ; calculate λ =lcm( p −1, q −1), where lcm represents the least common multiple; Pick a random integer g satisfy g The order is N 2 ; Calculate the private key μ Make μ × λ ≡1 mod N 2 ; The public key is ( N,g), the private key is ( λ , μ ).

[0041] In this embodiment, since the cloud server has the private key of the Paillier encryption algorithm, the encrypted feature ciphertext vector and ciphertext threshold vector are directly sent to the cloud server, and the cloud server will directly decrypt to obtain the original data information (i.e., historical resource transaction data), and will send the user's personal privacy data to the cloud server, which may easily lead to personal privacy leakage caused by the cloud server. Therefore, the following operations are performed: 2) Randomly generate a first reversible matrix, a second reversible matrix, a first triangular matrix, and a second triangular matrix.

[0042] The first reversible matrix and the second reversible matrix have the same dimension, both of which are m×m order, the elements on the diagonal of the first triangular matrix and the second triangular matrix are all 1, and the other elements are randomly generated, preferably 0.

[0043] 3) Generate the first diagonal matrix based on the ciphertext threshold vector, where the ciphertext threshold vector has m elements, one element corresponds to a feature threshold, first establish an m×m matrix in which each element is zero, and then use the m elements in the ciphertext threshold vector as the elements on the diagonal of the matrix, and then you can get a first diagonal matrix.

[0044] 4) generating a first threshold matrix based on the first reversible matrix, the first triangular matrix and the first diagonal matrix, generating a second threshold matrix based on the second reversible matrix, the second triangular matrix and the first diagonal matrix, and using the first threshold matrix and the second threshold matrix as the threshold encrypted data set; Among them, the calculation expression of the first threshold matrix is: , V 1 is the first threshold matrix, M 1 is the first reversible matrix, Q 1 is the first triangular matrix, D 1 is the first diagonal matrix; Among them, the calculation expression of the second threshold matrix is: , V 2 is the second threshold matrix, M 2 is the second reversible matrix, Q 2 is the second triangular matrix, D 1 is the first diagonal matrix.

[0045] 5) Generate the second diagonal matrix based on the characteristic ciphertext vector. Similarly, each historical resource transaction record has m elements in the characteristic ciphertext vector, and there are n characteristic ciphertext vectors in total. First, establish n m×m matrices. Each element in the n matrices is zero. The m elements in the i-th characteristic ciphertext vector are used as the elements on the diagonal of the i-th matrix, i=1,2,..,n. At this time, n second diagonal matrices can be obtained.

[0046] 6) Generate a second feature matrix based on the second diagonal matrix and the first reversible matrix, and use the second feature matrix as the feature encrypted data set; use the threshold encrypted data set and the feature encrypted data set as the encrypted second target data set.

[0047] Among them, the calculation expression of the second characteristic matrix is: ,in, is the i-th second characteristic matrix, M 1 is the first reversible matrix, is the i-th second diagonal matrix.

[0048] Step a30: sending the encrypted second target data set to a cloud server, and the cloud server constructs a transaction detection model according to the encrypted second target data set.

[0049] Specifically, the cloud server constructs a transaction detection model according to the encrypted second target data set, including: Step b10: The cloud server calculates the data set to be decrypted based on the feature encrypted data set and the threshold encrypted data set.

[0050] The cloud server multiplies the received feature encrypted data set and threshold encrypted data set by the first threshold matrix and the second feature matrix, that is, multiplies each element on the diagonal of the first threshold matrix with each corresponding element on the diagonal of the second feature matrix. Each transaction record can obtain m vectors to be decrypted, and the combination of vectors to be decrypted obtained from all transaction records constitutes an n×m output matrix to be decrypted, and all elements in the n×m output matrix to be decrypted are used as the output data set to be decrypted.

[0051] Step b20: The cloud server decrypts the data set to be decrypted based on the private key to obtain a decrypted data set.

[0052] The cloud server has a private key of the Paillier encryption algorithm. By using the private key to perform decryption operations on the data set to be decrypted, the decrypted data set can be calculated. The decrypted data set is also an n×m-order decryption matrix.

[0053] Step b30: Compare each element in the decrypted data set with a preset value to obtain a comparison result of each element in the decrypted data set.

[0054] In this embodiment, each element in the n×m order decryption matrix is ​​compared with a preset value, which is calculated based on the modulus. N Determine; when an element in the n×m order decryption matrix is ​​less than or equal to the preset value, a comparison result is obtained (the comparison result is 1 at this time); when an element in the n×m order decryption matrix is ​​greater than the preset value, a comparison result is obtained (the comparison result is 0 at this time); the comparison results of all elements can form an n×m order result matrix.

[0055] Step b40: Divide the comparison results of each element of the decrypted data set based on the single-layer decision tree and the feature labels to obtain classification labels.

[0056] In this embodiment, the steps of dividing the comparison results using a single-layer decision tree are as follows: one comparison result is divided using a single-layer decision tree. If the value of the comparison result corresponding to a historical resource transaction record is 1, the historical resource transaction record is divided into the right leaf node of the single-layer decision tree (the single-layer decision tree has a root node and two leaf nodes). If the value of the comparison result is 0, the historical resource transaction record is divided into the left leaf node of the single-layer decision tree. At this time, the division results of m single-layer decision trees can be obtained; in the process of division of each single-layer decision tree, the feature label of the historical resource transaction record is recorded.

[0057] Count the label values ​​on all left leaf nodes and the label values ​​on all right leaf nodes, and use the feature labels corresponding to the majority of features as the final classification labels.

[0058] Step b50: Determine a set of candidate base classifiers for a single-layer decision tree based on the classification labels, the comparison results of the elements of the decrypted data set, and the threshold encrypted data set.

[0059] In this embodiment, a single-layer decision tree is used as a base classifier, and the second eigenvector, the set of historical resource transaction records on the left leaf node of the single-layer decision tree, the set of historical resource transaction records on the right leaf node and the final classification label are used as the representation data set of the single-layer decision tree, and the representation data set can be used as a candidate base classifier set.

[0060] Step b60: Iteratively train the candidate base classifier set of the single-layer decision tree to obtain a number of base classifiers and coefficients corresponding to each base classifier.

[0061] In each iteration process, the class error rate of each base classifier in the candidate base classifier set is calculated, that is, the sum of the weight values ​​of the misclassified samples (a historical resource transaction record). At the same time, the single-layer decision tree with the lowest error rate is selected as the best base classifier for this round, and the coefficient corresponding to the best base classifier is calculated. The best base classifier is weighted and linearly added to the previous classifier to update the weight value of the next round of samples. After the iteration termination condition is met, several trained base classifiers and the coefficients corresponding to each base classifier can be obtained. Among them, the iteration condition is that the number of iterations reaches the preset number of times or the sum of the weight values ​​of the misclassified samples (a historical resource transaction record) is less than the preset weight value.

[0062] Step b70: constructing a transaction detection model based on a number of base classifiers and coefficients corresponding to each base classifier, that is, linearly weighting the number of base classifiers and coefficients corresponding to each base classifier to obtain the transaction detection model.

[0063] As a further optimization of this embodiment, in step S20, the step of encrypting the first feature vector is similar to the step of encrypting the second feature vector, and specifically includes the following steps: Step c10: Encrypt the first feature vector based on the public key of the encryption algorithm to obtain a first feature ciphertext vector; Step c20: generating a third diagonal matrix based on the first characteristic ciphertext vector; Step c30: Generate a first characteristic matrix based on the second reversible matrix and the third diagonal matrix, and use the first characteristic matrix as the first target data set.

[0064] As a further optimization of this embodiment, the transaction detection model deployed on the cloud server performs anomaly detection on the first target data set to obtain a detection result, including: Step d10: generating a classification label for each element in the first target data set based on the transaction detection model; Step d10: Determine the detection result based on the classification label of each element in the first target data set and the coefficient corresponding to each base classifier.

[0065] In this embodiment, after obtaining the real-time resource transaction record, the real-time resource transaction record is converted into a first target data set, which is a matrix obtained by multiplying the second reversible matrix by the third diagonal matrix. The cloud server uses the same decryption algorithm to decrypt the first target data set to obtain a decryption vector corresponding to the real-time resource transaction record (i.e., a decryption vector formed by decrypting all elements on the diagonal of the first target data set).

[0066] Each element in the decrypted vector is then compared with a preset value, which is calculated based on the modulus NDetermine that when an element in the decryption vector is less than or equal to a preset value, a comparison result is obtained (the comparison result is 1 at this time); when an element in the decryption vector is greater than the preset value, a comparison result is obtained (the comparison result is 0 at this time); the comparison results of all elements in the decryption vector can constitute a result vector.

[0067] The base classifier and classification label corresponding to each element in the result vector are searched in all base classifiers, and the values ​​of all classification labels are linearly summed with the coefficients of the corresponding base classifiers to obtain the detection result.

[0068] In this embodiment, since the first reversible matrix and the second reversible matrix are randomly generated, the first characteristic matrix needs to be restored in the decryption stage, and the first reversible matrix and the second reversible matrix cannot be solved, so that the original information of the real-time resource transaction record will not be transmitted to the cloud server, thereby improving the security of data transmission. Therefore, the present invention can protect data privacy while ensuring the accuracy and efficiency of anomaly detection.

[0069] Step S40: Obtain the detection results of the cloud server, generate a detection report based on the detection results, and visualize the detection report.

[0070] In this embodiment, the transaction server can check whether there is an abnormality in the current real-time resource transaction record, and the transaction server can also push the detection result to the user of the transaction terminal; it can also be saved only on the transaction server and not pushed to the transaction terminal.

[0071] The present invention constructs a transaction detection model through historical resource transaction records, which contain a variety of sample data; then the transaction detection model is used to perform anomaly detection on real-time resource transaction records, which can improve the accuracy and scope of anomaly detection; and in the detection process, the real-time resource transaction records are encrypted, which can effectively ensure that the user's privacy will not be leaked.

[0072] Embodiment 2 Figure 2 is a block diagram of a data automation resource transaction management system provided by one embodiment of the present invention, such as Figure 2 As shown, this embodiment provides a data automation resource transaction management system, the system is used to implement the data automation resource transaction management method of embodiment 1, and the system includes: Data acquisition module, used to obtain real-time resource transaction records; A data processing module, used for preprocessing and encrypting the real-time resource transaction records to obtain a first target data set; A data detection module is used to upload the first target data set to the cloud server, and the transaction detection model deployed on the cloud server performs anomaly detection on the first target data set to obtain a detection result; wherein the transaction detection model is constructed based on historical resource transaction records; The detection display module is used to obtain the detection results of the cloud server, generate a detection report based on the detection results, and visually display the detection report.

[0073] This embodiment also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the above-mentioned data automated resource transaction management method is implemented.

[0074] The present invention constructs a transaction detection model through historical resource transaction records, which contain a variety of sample data; and then uses the transaction detection model to perform anomaly detection on real-time resource transaction records, which can improve the accuracy and range of anomaly detection; and in the detection process, the real-time resource transaction records are encrypted, which can effectively ensure that the user's privacy will not be leaked. Therefore, the present invention can protect data privacy while ensuring the accuracy and efficiency of anomaly detection.

[0075] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0076] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0077] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A data automation resource transaction management method, characterized in that: The method comprises: Obtain real-time resource transaction records; Preprocessing and encrypting the real-time resource transaction records to obtain a first target data set; The first target data set is uploaded to the cloud server, and a transaction detection model deployed on the cloud server performs anomaly detection on the first target data set to obtain a detection result; wherein the transaction detection model is constructed based on historical resource transaction records; Obtain the test results of the cloud server, generate a test report based on the test results, and visualize the test report.

2. The data automation resource transaction management method according to claim 1, characterized in that: The method further includes: constructing a transaction detection model, including: Acquire historical resource transaction records, pre-process the historical resource transaction records, and obtain a second target data set; encrypting the second target data set to obtain an encrypted second target data set; The encrypted second target data set is sent to a cloud server, and the cloud server constructs a transaction detection model according to the encrypted second target data set.

3. The data automation resource transaction management method according to claim 2, characterized in that: Preprocess the historical resource transaction records to obtain the second target data set, including: Extract features from historical resource transaction records to obtain a second feature vector; Record the label of the historical resource transaction record to obtain the characteristic label, wherein the historical resource transaction record contains the label of the abnormal transaction; The second eigenvector is combined with the feature label to obtain a transaction record data matrix; A preset number of columns are extracted from the transaction record data matrix, feature thresholds are calculated based on all elements in each extracted column, a threshold vector is constructed based on the feature thresholds corresponding to all extracted columns, and the threshold vector, the second feature vector and the feature label are used as the second target data set.

4. The data automation resource transaction management method according to claim 3, characterized in that: The second target data set is encrypted to obtain an encrypted second target data set, including: The second feature vector and the threshold vector are encrypted respectively based on the public key of the encryption algorithm to obtain a feature ciphertext vector corresponding to the second feature vector and a ciphertext threshold vector corresponding to the threshold vector; Randomly generate a first reversible matrix, a second reversible matrix, a first triangular matrix, and a second triangular matrix; Generate a first diagonal matrix based on the ciphertext threshold vector; Generate a first threshold matrix based on the first reversible matrix, the first triangular matrix and the first diagonal matrix, generate a second threshold matrix based on the second reversible matrix, the second triangular matrix and the first diagonal matrix, and use the first threshold matrix and the second threshold matrix as threshold encrypted data sets; Generate a second diagonal matrix based on the characteristic ciphertext vector; A second characteristic matrix is ​​generated based on the second diagonal matrix and the first reversible matrix, and the second characteristic matrix is ​​used as a characteristic encrypted data set; and the threshold encrypted data set and the characteristic encrypted data set are used as the encrypted second target data set.

5. The data automation resource transaction management method according to claim 4, characterized in that: The cloud server constructs a transaction detection model according to the encrypted second target data set, including: The cloud server calculates the data set to be decrypted based on the feature encrypted data set and the threshold encrypted data set; The cloud server decrypts the data set to be decrypted based on the private key to obtain a decrypted data set; Compare each element in the decrypted data set with a preset value to obtain a comparison result of each element of the decrypted data set; Based on the single-layer decision tree and feature labels, the comparison results of each element of the decrypted data set are divided to obtain classification labels; Determine a set of candidate base classifiers of a single-layer decision tree based on the classification labels, the comparison results of each element of the decrypted data set, and the threshold encrypted data set; Iteratively train the candidate base classifier set of the single-layer decision tree to obtain a number of base classifiers and coefficients corresponding to each base classifier; A transaction detection model is constructed based on several base classifiers and coefficients corresponding to each base classifier.

6. The data automation resource transaction management method according to claim 5, characterized in that: The transaction detection model deployed on the cloud server performs anomaly detection on the first target data set to obtain detection results, including: Generate a classification label for each element in the first target data set based on the transaction detection model; A detection result is determined based on the classification label of each element in the first target data set and the coefficient corresponding to each base classifier.

7. The data automation resource transaction management method according to claim 4, characterized in that: The real-time resource transaction records are preprocessed and encrypted to obtain a first target data set, including: Extracting features from real-time resource transaction records to obtain a first feature vector; The first feature vector is encrypted to obtain a first target data set.

8. The data automation resource transaction management method according to claim 7, characterized in that: The first feature vector is encrypted to obtain a first target data set, including: Encrypting the first feature vector based on a public key of an encryption algorithm to obtain a first feature ciphertext vector; Generate a third diagonal matrix based on the first characteristic ciphertext vector; A first characteristic matrix is ​​generated based on the second reversible matrix and the third diagonal matrix, and the first characteristic matrix is ​​used as a first target data set.

9. A data automation resource transaction management system, the system being used to implement the data automation resource transaction management method according to any one of claims 1 to 8, characterized in that: The system comprises: Data acquisition module, used to obtain real-time resource transaction records; A data processing module, used for preprocessing and encrypting the real-time resource transaction records to obtain a first target data set; A data detection module, used for uploading the first target data set to the cloud server, and performing anomaly detection on the first target data set by a transaction detection model deployed on the cloud server to obtain a detection result; wherein the transaction detection model is constructed based on historical resource transaction records; The detection display module is used to obtain the detection results of the cloud server, generate a detection report based on the detection results, and visually display the detection report.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the data automation resource transaction management method described in any one of claims 1 to 8 is implemented.