Transaction risk assessment method and device based on artificial intelligence
Through the transaction risk assessment method based on artificial intelligence, the data characteristics of buyers and sellers are used to build a risk model and calculate transaction risk values, which solves the transaction risk problem caused by the inability to effectively identify and prevent theft of accounts and passwords in the existing technology, and improves the security of transactions and the accuracy of risk assessment.
Patent Information
- Application Number
- CN202510042374.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-10
- Publication Date
- 2025-05-09
AI Technical Summary
The existing technology relies on the matching of bank card accounts and passwords in transaction risk assessment, and cannot effectively identify and prevent transaction risks caused by accounts and passwords stolen through virus software and other means.
Using an artificial intelligence-based transaction risk assessment method, by obtaining current and historical data information of buyers and sellers, extracting data statistical features and transaction risk characteristics, building a characteristic data matrix, input it into the transaction risk identification model, calculating the transaction risk value, and prohibiting transactions when the risk value exceeds the preset threshold.
It improves the accuracy and security of transaction risk assessment, and can automatically calculate transaction risk values before transactions occur, preventing transaction risks caused by theft of account and passwords.
Smart Images
Figure CN119963197A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of data processing technology, and in particular to a transaction risk assessment method and device based on artificial intelligence. Background Art
[0002] With the continuous development of the Internet and the continuous progress of the economy, the daily services provided to people have shifted from offline to online. People use online software to shop, take taxis, wash cars, order meals, hire hourly workers, etc. When conducting these online activities, people need to submit online orders and make online payments before or after completing the corresponding services. In order to protect the interests of consumers, online software will assess the transaction risks before consumers make online payments, and determine whether the consumer's online software account has been stolen based on the assessment results.
[0003] In the prior art, when evaluating a transaction, it is usually determined by determining whether the bank card account number and the input password match. If they match, the transaction can be executed; if they do not match, the transaction is terminated. However, as criminals become more sophisticated, they can obtain users' accounts and passwords through various means, such as by implanting virus software on users' mobile phones or computers, so that bank deposits are still often stolen in real life. Based on this, how to improve the accuracy of transaction risk assessment has become one of the technical problems that technicians in this field need to solve urgently. Summary of the invention
[0004] In view of this, the present application provides a transaction risk assessment method, device, electronic device and storage medium based on artificial intelligence, which are used to improve the accuracy of personal credit assessment.
[0005] In a first aspect, an embodiment of the present application provides a transaction risk assessment method based on artificial intelligence, the method comprising:
[0006] Acquire the transaction data information of the current transaction to be traded, wherein the transaction data information includes the current buyer data information, the current seller data information and the current transaction commodity data information;
[0007] Acquire the buyer's historical data information corresponding to the current buyer's data information, and the seller's historical data information corresponding to the current seller's data information;
[0008] Extracting buyer data statistical features and buyer transaction risk features based on the current buyer data information and the buyer historical data information; and extracting seller data statistical features and seller transaction risk features based on the current seller data information and the seller historical data information;
[0009] Determine a first characteristic data matrix according to the buyer's data statistical characteristics and the buyer's transaction risk characteristics, determine a second characteristic data matrix according to the seller's data statistical characteristics and the seller's transaction risk characteristics, and determine a third characteristic data matrix according to the current transaction commodity data information;
[0010] Determining the transaction risk value of the current transaction to be traded according to the first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix;
[0011] When the risk value of the transaction currently to be traded is greater than the preset risk value, the transaction currently to be traded is prohibited.
[0012] In an optional embodiment provided by the present invention, the extracting buyer data statistical features and buyer transaction risk features according to the current buyer data information and the buyer historical data information includes:
[0013] Extracting data from the current buyer data information and the buyer historical data information respectively to obtain buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount;
[0014] Determine the buyer data statistical characteristics based on the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount;
[0015] The keyword matching result is obtained by performing keyword matching on the buyer transaction type, buyer device basic information, and buyer device usage information of the current buyer data information through the keywords in the first mapping table, wherein the first mapping table stores a plurality of keywords and their corresponding risk values;
[0016] The buyer's transaction risk characteristics are determined based on the keyword matching results and their corresponding risk values.
[0017] In an optional embodiment provided by the present invention, determining the buyer data statistical features according to the obtained buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount includes:
[0018] Matching the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount respectively corresponding to the current buyer data information and the buyer historical data information to obtain a data item matching result, and obtaining first weight values respectively corresponding to the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount according to the data item matching result;
[0019] Matching the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount of the current buyer data information through the weight information table to determine second weight values corresponding to the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount, respectively;
[0020] The buyer data statistical characteristics are determined according to the buyer transaction type, buyer device basic information, buyer device usage information, buyer transaction amount and their corresponding first weight values and second weight values respectively.
[0021] In an optional embodiment provided by the present invention, the extracting the seller data statistical features and the seller transaction risk features according to the current seller data information and the seller historical data information includes:
[0022] Extracting data from the current seller data information and the seller historical data information respectively to obtain seller transaction records, seller payment records, and seller transaction methods;
[0023] Determine the seller data statistical features according to the seller transaction records, seller payment records and seller transaction methods respectively corresponding to the current seller data information and the seller historical data information;
[0024] Convert the seller transaction record, seller payment record and seller transaction method of the current seller data information into a first feature vector, and convert the seller transaction record, seller payment record and seller transaction method of the seller historical data information into a second feature vector;
[0025] Inputting the first feature vector and the second feature vector into a seller transaction risk identification model to obtain risk weight values corresponding to the seller transaction record, seller payment record, and seller transaction method of the current seller data information;
[0026] The seller data statistical characteristics are determined based on the seller transaction records, seller payment records and seller transaction methods of the current seller data information and their corresponding risk weight values.
[0027] In an optional embodiment provided by the present invention, the first feature vector and the second feature vector are input into the seller transaction risk identification model to obtain the risk weight values corresponding to the seller transaction record, the seller payment record and the seller transaction method of the current seller data information, respectively, including:
[0028] Inputting the first feature vector and the second feature vector into a seller's transaction risk identification model to obtain a transaction risk prediction value;
[0029] According to the importance of corresponding data items in the first feature vector and the second feature vector to obtaining the transaction risk prediction value, the risk weight values corresponding to the seller transaction record, the seller payment record and the seller transaction method of the current seller data information are determined.
[0030] In an optional embodiment provided by the present invention, determining the transaction risk value of the current transaction to be traded according to the first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix includes:
[0031] The first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix are input into the current transaction risk identification model to obtain the transaction risk value of the current transaction.
[0032] In an optional embodiment provided by the present invention, the first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix are input into the current transaction risk identification model to obtain the transaction risk value of the current transaction, including:
[0033] Inputting the first feature data matrix, the second feature data matrix and the third feature data matrix into the current transaction risk identification model, and obtaining data convolution features corresponding to the first feature data matrix, the second feature data matrix and the third feature data matrix respectively through different convolution layers in the current transaction risk identification model;
[0034] Input any two data convolution features into the corresponding prediction module to obtain the corresponding prediction value;
[0035] The transaction risk value of the current transaction to be traded is obtained according to the obtained prediction value.
[0036] In an optional embodiment provided by the present invention, obtaining the transaction risk value of the current pending transaction according to the obtained prediction value includes:
[0037] The transaction risk value of the current transaction to be traded is calculated based on each prediction value and its corresponding prediction probability value, as well as the accuracy of each prediction module.
[0038] In an optional embodiment provided by the present invention, the training process of the current transaction risk identification model is:
[0039] Obtaining sample data for training the current transaction risk identification model and marking risk values; the sample data includes a first sample data matrix, a second sample data matrix, and a third sample data matrix;
[0040] Inputting the first sample data matrix, the second sample data matrix, and the third sample data matrix into the current transaction risk identification model to obtain a first prediction value, a second prediction value, and a third prediction value;
[0041] Calculating a final prediction value according to the first prediction value, the second prediction value and the third prediction value;
[0042] The loss value is calculated by using the final predicted value and the marked risk value. When the loss value is less than the target loss value, the training of the current transaction risk identification model is completed.
[0043] In a second aspect, an embodiment of the present application further provides a transaction risk assessment device based on artificial intelligence, the device comprising:
[0044] An acquisition module is used to acquire the transaction data information of the current transaction to be traded, wherein the transaction data information includes the current buyer data information, the current seller data information and the current transaction commodity data information;
[0045] The acquisition module is further used to acquire the buyer's historical data information corresponding to the current buyer's data information, and the seller's historical data information corresponding to the current seller's data information;
[0046] An extraction module, configured to extract buyer data statistical features and buyer transaction risk features based on the current buyer data information and the buyer historical data information; and to extract seller data statistical features and seller transaction risk features based on the current seller data information and the seller historical data information;
[0047] A determination module, configured to determine a first characteristic data matrix according to the buyer's data statistical characteristics and the buyer's transaction risk characteristics, determine a second characteristic data matrix according to the seller's data statistical characteristics and the seller's transaction risk characteristics, and determine a third characteristic data matrix according to the current transaction commodity data information;
[0048] The determination module is further used to determine the transaction risk value of the current pending transaction based on the first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix;
[0049] The prohibition module is used to prohibit the transaction currently to be traded when the risk value of the transaction currently to be traded is greater than the preset risk value.
[0050] In a third aspect, an embodiment of the present application further provides an electronic device, comprising: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate through the bus, and the machine-readable instructions are executed by the processor to execute the steps of the artificial intelligence-based transaction risk assessment method of the first aspect.
[0051] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the artificial intelligence-based transaction risk assessment method in the first aspect are executed.
[0052] An artificial intelligence-based transaction risk assessment method and device provided in an embodiment of the present application first obtain transaction data information of a current transaction to be traded, wherein the transaction data information includes current buyer data information, current seller data information, and current transaction commodity data information; then obtain buyer historical data information corresponding to the current buyer data information, and seller historical data information corresponding to the current seller data information; extract buyer data statistical features and buyer transaction risk features based on the current buyer data information and buyer historical data information; and extract seller data statistical features and seller transaction risk features based on the current seller data information and seller historical data information; determine a first feature data matrix based on the buyer data statistical features and buyer transaction risk features, determine a second feature data matrix based on the seller data statistical features and seller transaction risk features, and determine a third feature data matrix based on the current transaction commodity data information; finally, determine a transaction risk value of the current transaction to be traded based on the first feature data matrix, the second feature data matrix, and the third feature data matrix; when the transaction risk value of the current transaction to be traded is greater than a preset risk value, prohibit the current transaction to be traded. Compared with the prior art that evaluates transactions by whether the bank card account number and the input password match, the present application can automatically calculate the transaction risk value of the current pending transaction through the buyer's data information and the seller's data information before the actual transaction occurs, and when the transaction risk value is greater than the preset risk value, the current pending transaction is prohibited. Therefore, the present application can improve the security of transactions and the accuracy of transaction risk assessment.
[0053] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0054] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.
[0055] Figure 1 A flowchart of a transaction risk assessment method based on artificial intelligence provided in an embodiment of the present application is shown;
[0056] Figure 2 A structural block diagram of a transaction risk assessment device based on artificial intelligence provided in an embodiment of the present application is shown;
[0057] Figure 3 A schematic diagram of a computer device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION
[0058] The terms "first", "second", "third", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish different objects rather than to limit a specific order.
[0059] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a concrete way for easy understanding.
[0060] In the description of this application, unless otherwise specified, " / " indicates that the objects associated before and after are in an "or" relationship, for example, A / B can represent A or B; "and / or" in this application is only a description of the association relationship of associated objects, indicating that there can be three relationships, for example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. In addition, in the description of this application, unless otherwise specified, "multiple" refers to two or more than two. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or multiple.
[0061] In the embodiments of the present application, at least one can also be described as one or more, and multiple can be two, three, four or more, which is not limited in the present application.
[0062] like Figure 1 As shown, the embodiment of the present application provides a transaction risk assessment method based on artificial intelligence. The transaction risk assessment method based on artificial intelligence provided by the present application may include:
[0063] S10. Acquire transaction data information of a current transaction, wherein the transaction data information includes current buyer data information, current seller data information, and current transaction commodity data information.
[0064] Among them, the current buyer data information includes the buyer's account, buyer's user ID, buyer's name, transaction method, buyer's location, and transaction amount, etc.; the current seller data information includes the seller's account, seller's user ID, seller's name, transaction amount, seller's location, etc.; the current transaction product data information includes product type, delivery time, transaction type, etc., and this embodiment does not make specific limitations on this.
[0065] S20. Obtain the buyer's historical data information corresponding to the current buyer's data information, and the seller's historical data information corresponding to the current seller's data information.
[0066] Specifically, this embodiment can obtain the buyer's historical data information corresponding to the current buyer's data information through the buyer's user ID, and obtain the seller's historical data information corresponding to the current seller's data information through the seller's user ID. Among them, this embodiment can obtain historical data information within a certain period of time, such as obtaining the buyer's historical data information and the seller's historical data information for the last one or three months.
[0067] S30. Extracting buyer data statistical features and buyer transaction risk features based on current buyer data information and buyer historical data information; and extracting seller data statistical features and seller transaction risk features based on current seller data information and seller historical data information.
[0068] In an optional embodiment provided by the present invention, the extracting buyer data statistical features and buyer transaction risk features according to the current buyer data information and the buyer historical data information includes:
[0069] S3011. Extract the current buyer data information and the buyer historical data information respectively to obtain the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount.
[0070] Among them, the buyer's transaction type is classified according to the different nature of the transaction, including purchase expenditure, transfer expenditure, payment, refund, recharge, withdrawal, and remittance; the buyer's device basic information includes the device name, model, serial number, specifications, manufacturer, production date, purchase date, device number, etc.; the buyer's device usage information includes the account registration record, login duration, communication record, and recent contacts of the APP in the device, etc. This embodiment does not make specific limitations on this.
[0071] S3012. Determine the buyer data statistical characteristics based on the obtained buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount.
[0072] In an optional embodiment, the buyer data statistical characteristics are determined based on the obtained buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount, including: matching the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount corresponding to the current buyer data information and the buyer historical data information to obtain data item matching results, and obtaining first weight values corresponding to the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount according to the data item matching results; matching the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount of the current buyer data information through a weight information table to determine the second weight values corresponding to the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount; determining the buyer data statistical characteristics based on the buyer transaction type, buyer device basic information, buyer device usage information, buyer transaction amount, and their corresponding first weight values and second weight values.
[0073] Specifically, the buyer transaction type of the current buyer data information and the buyer historical data information is matched to obtain a transaction type matching result, the buyer device basic information is matched to obtain a device basic information matching result, the buyer device usage information is matched to obtain a device usage information matching result, and the buyer transaction amount is matched to obtain a transaction amount matching result. Then, according to the matching results, the first weight values corresponding to the buyer transaction type, the buyer device basic information, the buyer device usage information, and the buyer transaction amount are obtained. For example, if the transaction type matching result shows that the current buyer transaction type has never appeared in its historical data, such as the current buyer transaction type is displayed as a virtual game transaction, and the transaction type does not appear in the buyer's historical transaction type, then the first weight value corresponding to the data item can be determined to be 1; if the basic information matching result shows that the buyer has changed the device used, then the first weight value corresponding to the data item can be determined to be 1. That is, in this embodiment, the weight value can be determined by the number of data mismatches in the matching result. The more mismatches there are, the larger the corresponding first weight value is.
[0074] In addition, this embodiment also needs to match the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount of the current buyer data information through the weight information table to determine the second weight values corresponding to the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount. Among them, the weight information table stores the weight values corresponding to various transaction types, device basic information, device usage information, and buyer transaction amounts. If the buyer device usage information of the current buyer data information includes the use of a dangerous or abnormal APP, or logging into a dangerous website, the corresponding second weight value in the weight information table.
[0075] In this embodiment, after obtaining the first weight value and the second weight value, a weighted calculation or an average value is performed on the first weight value and the second weight value to obtain weight values corresponding to the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount, respectively. Then, the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount are respectively converted into feature vectors, and the feature vectors are multiplied by the corresponding weight values to obtain the buyer data statistical features.
[0076] S3013. Perform keyword matching on the buyer transaction type, buyer device basic information, and buyer device usage information of the current buyer data information using the keywords in the first mapping table to obtain a keyword matching result.
[0077] The first mapping table stores multiple keywords and their corresponding risk values. Specifically, firstly, keyword extraction is performed on the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount to obtain multiple keywords, and then the keywords are matched with the keywords in the first mapping table to obtain keyword matching results, and the corresponding risk values in the keyword matching results are obtained.
[0078] For example, the first mapping table contains the following keywords: application A, web link B, virtual transaction, and the buyer's transaction address is a certain country. Then, through keyword matching, if the keywords application A, web link B, and virtual transaction match the keywords in the first mapping table, the risk value corresponding to the matching keywords is obtained, such as the risk value of application A is 1, the risk value of web link B is 2, and the risk value of virtual transaction is 3.
[0079] S3014. Determine the buyer's transaction risk characteristics based on the keyword matching results and their corresponding risk values.
[0080] Specifically, this embodiment converts the matched keywords into keyword feature vectors, and then multiplies the keyword feature vectors and the corresponding risk values to obtain the buyer's transaction risk characteristics.
[0081] In an optional embodiment provided by the present invention, the extracting the seller data statistical features and the seller transaction risk features according to the current seller data information and the seller historical data information includes:
[0082] S3021. Extract the current seller data information and the seller historical data information respectively to obtain the seller transaction record, the seller payment record and the seller transaction method.
[0083] S3022. Determine the statistical characteristics of the seller data according to the seller transaction records, seller payment records and seller transaction methods corresponding to the current seller data information and the seller historical data information respectively.
[0084] S3023. Convert the seller transaction records, seller payment records and seller transaction methods of the current seller data information into a first feature vector, and convert the seller transaction records, seller payment records and seller transaction methods of the seller historical data information into a second feature vector.
[0085] S3024. Input the first feature vector and the second feature vector into a seller transaction risk identification model to obtain risk weight values corresponding to the seller transaction record, seller payment record, and seller transaction method of the current seller data information.
[0086] Specifically, the step of inputting the first feature vector and the second feature vector into the seller transaction risk identification model to obtain the risk weight values respectively corresponding to the seller transaction records, seller payment records and seller transaction methods of the current seller data information includes: inputting the first feature vector and the second feature vector into the seller transaction risk identification model to obtain a transaction risk prediction value; determining the risk weight values respectively corresponding to the seller transaction records, seller payment records and seller transaction methods of the current seller data information according to the importance of the corresponding data items in the first feature vector and the second feature vector to obtaining the transaction risk prediction value.
[0087] The transaction risk identification model is a pre-trained neural network model, which converts sample data into a first sample feature vector and a second sample feature vector, and the sample label is the labeled transaction risk value. The first sample feature vector and the second sample feature vector are obtained in the same way as the first feature vector and the second feature vector, and this implementation will not be repeated here.
[0088] S3025. Determine the statistical characteristics of the seller data based on the seller transaction records, seller payment records, and seller transaction methods of the current seller data information and their corresponding risk weight values.
[0089] S40. Determine a first characteristic data matrix based on the buyer's data statistical characteristics and the buyer's transaction risk characteristics, determine a second characteristic data matrix based on the seller's data statistical characteristics and the seller's transaction risk characteristics, and determine a third characteristic data matrix based on the current transaction commodity data information.
[0090] Specifically, the buyer's data statistical features and the buyer's transaction risk features are directly combined to obtain a first feature data matrix, the seller's data statistical features and the seller's transaction risk features are combined to obtain a second feature data matrix, keywords are extracted from the previous transaction product data information, and then the third feature data matrix is determined based on the keyword vectors converted from the extracted keywords.
[0091] S50: Determine the transaction risk value of the current transaction to be traded according to the first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix.
[0092] In an optional embodiment provided by the present invention, determining the transaction risk value of the current transaction to be traded based on the first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix includes: inputting the first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix into a current transaction risk identification model to obtain the transaction risk value of the current transaction to be traded.
[0093] Specifically, the first characteristic data matrix, the second characteristic data matrix, and the third characteristic data matrix are input into the current transaction risk identification model to obtain the transaction risk value of the current transaction to be traded, including:
[0094] S501. Input the first feature data matrix, the second feature data matrix and the third feature data matrix into the current transaction risk identification model, and obtain the data convolution features corresponding to the first feature data matrix, the second feature data matrix and the third feature data matrix respectively through different convolution layers in the current transaction risk identification model.
[0095] Specifically, the current transaction risk identification model includes a first convolution layer, a second convolution layer and a third convolution layer. The present application inputs the first feature data matrix into the first convolution layer to obtain the first data convolution feature, inputs the second feature data matrix into the second convolution layer to obtain the second data convolution feature, and inputs the third feature data matrix into the third convolution layer to obtain the third data convolution feature.
[0096] In an optional embodiment provided by the present invention, the training process of the current transaction risk identification model is:
[0097] S5011. Obtain sample data for training the current transaction risk identification model and annotate risk values.
[0098] The sample data includes a first sample data matrix, a second sample data matrix, and a third sample data matrix. The first sample data matrix, the second sample data matrix, and the third sample data matrix are determined in the same manner as the first feature data matrix, the second feature data matrix, and the third feature data matrix, and are not described in detail in this embodiment.
[0099] S5012. Input the first sample data matrix, the second sample data matrix, and the third sample data matrix into the current transaction risk identification model to obtain a first prediction value, a second prediction value, and a third prediction value.
[0100] That is, after the first sample data matrix, the second sample data matrix, and the third sample data matrix are input into the current transaction risk identification model, the data convolution features corresponding to the first sample data matrix, the second sample data matrix, and the third sample data matrix are respectively extracted through the first convolution layer, the second convolution layer, and the third convolution layer. Then, the first prediction value, the second prediction value, and the third prediction value are obtained through three prediction modules.
[0101] S5013. Calculate a final prediction value according to the first prediction value, the second prediction value, and the third prediction value.
[0102] Specifically, the first prediction value, the second prediction value, and the third prediction value are weighted to obtain the final prediction value. The weight values corresponding to the first prediction value, the second prediction value, and the third prediction value are determined according to the prediction accuracy rates corresponding to the first prediction module, the second prediction module, and the third prediction module, respectively. The higher the accuracy rate, the greater the corresponding weight value.
[0103] S5014. Calculate the loss value using the final predicted value and the marked risk value. When the loss value is less than the target loss value, the training of the current transaction risk identification model is completed.
[0104] S502: Input any two data convolution features into the corresponding prediction module to obtain the corresponding prediction value.
[0105] Specifically, the first data convolution feature and the second data convolution feature are input into the first prediction module to obtain a first prediction value, the first data convolution feature and the third data convolution feature are input into the second prediction module to obtain a second prediction value, and the second data convolution feature and the third data convolution feature are input into the third prediction module to obtain a third prediction value.
[0106] S503: Obtain the transaction risk value of the current transaction to be traded according to the obtained prediction value.
[0107] Wherein, obtaining the transaction risk value of the current transaction to be traded according to the obtained prediction value includes: obtaining the transaction risk value of the current transaction to be traded by calculating according to each prediction value and its corresponding prediction probability value, and the accuracy of each prediction module.
[0108] Specifically, in this embodiment, the transaction risk value of the current transaction to be traded can be calculated by the following formula:
[0109] F=[(f1x1+f2x2+f3x3) / 3+(f1y1+f2y2+f3y3) / 3] / 2
[0110] Among them, f1 is the first prediction value obtained by the first prediction module, f2 is the second prediction value obtained by the second prediction module, and f3 is the second prediction value obtained by the first prediction module; x1 is the prediction probability value corresponding to the first prediction value, x2 is the prediction probability value corresponding to the second prediction value, and x3 is the prediction probability value corresponding to the third prediction value; y1 is the accuracy of the first prediction module, f2 is the accuracy of the second prediction module, and y3 is the accuracy of the third prediction module.
[0111] S60: When the risk value of the transaction currently to be traded is greater than the preset risk value, prohibiting the transaction currently to be traded.
[0112] Among them, the preset risk value is a value set according to actual needs.
[0113] An artificial intelligence-based transaction risk assessment method provided in an embodiment of the present application first obtains transaction data information of a current transaction to be traded, wherein the transaction data information includes current buyer data information, current seller data information, and current transaction commodity data information; then obtains buyer historical data information corresponding to the current buyer data information, and seller historical data information corresponding to the current seller data information; extracts buyer data statistical features and buyer transaction risk features based on the current buyer data information and buyer historical data information; and extracts seller data statistical features and seller transaction risk features based on the current seller data information and seller historical data information; determines a first feature data matrix based on the buyer data statistical features and buyer transaction risk features, determines a second feature data matrix based on the seller data statistical features and seller transaction risk features, and determines a third feature data matrix based on the current transaction commodity data information; finally, determines a transaction risk value of the current transaction to be traded based on the first feature data matrix, the second feature data matrix, and the third feature data matrix; when the transaction risk value of the current transaction to be traded is greater than a preset risk value, prohibits the current transaction to be traded. Compared with the prior art that evaluates transactions by whether the bank card account number and the input password match, the present application can automatically calculate the transaction risk value of the current pending transaction through the buyer's data information and the seller's data information before the actual transaction occurs, and when the transaction risk value is greater than the preset risk value, the current pending transaction is prohibited. Therefore, the present application can improve the security of transactions and the accuracy of transaction risk assessment.
[0114] In the case of dividing each functional module into corresponding functional modules, Figure 2 A possible schematic diagram of the composition of the transaction risk assessment device based on artificial intelligence involved in the above and embodiments is shown, Figure 2 As shown, the transaction risk assessment device based on artificial intelligence may include:
[0115] The acquisition module 21 is used to acquire the transaction data information of the current transaction to be traded, wherein the transaction data information includes the current buyer data information, the current seller data information and the current transaction commodity data information;
[0116] The acquisition module 21 is further used to acquire the buyer's historical data information corresponding to the current buyer's data information, and the seller's historical data information corresponding to the current seller's data information;
[0117] An extraction module 22 is used to extract buyer data statistical features and buyer transaction risk features based on the current buyer data information and the buyer historical data information; and to extract seller data statistical features and seller transaction risk features based on the current seller data information and the seller historical data information;
[0118] A determination module 23, configured to determine a first characteristic data matrix according to the buyer's data statistical characteristics and the buyer's transaction risk characteristics, determine a second characteristic data matrix according to the seller's data statistical characteristics and the seller's transaction risk characteristics, and determine a third characteristic data matrix according to the current transaction commodity data information;
[0119] The determination module 23 is further used to determine the transaction risk value of the current transaction to be traded according to the first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix;
[0120] The prohibition module 24 is used to prohibit the transaction currently to be traded when the risk value of the transaction currently to be traded is greater than the preset risk value.
[0121] In an optional embodiment provided by the present invention, the extraction module 22 is specifically used for:
[0122] Extracting data from the current buyer data information and the buyer historical data information respectively to obtain buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount;
[0123] Determine the buyer data statistical characteristics based on the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount;
[0124] The keyword matching result is obtained by performing keyword matching on the buyer transaction type, buyer device basic information, and buyer device usage information of the current buyer data information through the keywords in the first mapping table, wherein the first mapping table stores a plurality of keywords and their corresponding risk values;
[0125] The buyer's transaction risk characteristics are determined based on the keyword matching results and their corresponding risk values.
[0126] In an optional embodiment provided by the present invention, the extraction module 22 is specifically used for:
[0127] Matching the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount respectively corresponding to the current buyer data information and the buyer historical data information to obtain a data item matching result, and obtaining first weight values respectively corresponding to the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount according to the data item matching result;
[0128] Matching the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount of the current buyer data information through the weight information table to determine second weight values corresponding to the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount, respectively;
[0129] The buyer data statistical characteristics are determined according to the buyer transaction type, buyer device basic information, buyer device usage information, buyer transaction amount and their corresponding first weight values and second weight values respectively.
[0130] In an optional embodiment provided by the present invention, the extraction module 22 is specifically used for:
[0131] Extracting data from the current seller data information and the seller historical data information respectively to obtain seller transaction records, seller payment records, and seller transaction methods;
[0132] Determine the seller data statistical features according to the seller transaction records, seller payment records and seller transaction methods respectively corresponding to the current seller data information and the seller historical data information;
[0133] Convert the seller transaction record, seller payment record and seller transaction method of the current seller data information into a first feature vector, and convert the seller transaction record, seller payment record and seller transaction method of the seller historical data information into a second feature vector;
[0134] Inputting the first feature vector and the second feature vector into a seller transaction risk identification model to obtain risk weight values corresponding to the seller transaction record, seller payment record, and seller transaction method of the current seller data information;
[0135] The seller data statistical characteristics are determined based on the seller transaction records, seller payment records and seller transaction methods of the current seller data information and their corresponding risk weight values.
[0136] In an optional embodiment provided by the present invention, the extraction module 22 is specifically used for:
[0137] Inputting the first feature vector and the second feature vector into a seller's transaction risk identification model to obtain a transaction risk prediction value;
[0138] According to the importance of corresponding data items in the first feature vector and the second feature vector to obtaining the transaction risk prediction value, the risk weight values corresponding to the seller transaction record, the seller payment record and the seller transaction method of the current seller data information are determined.
[0139] In an optional embodiment provided by the present invention, the determination module 23 is specifically configured to:
[0140] The first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix are input into the current transaction risk identification model to obtain the transaction risk value of the current transaction.
[0141] In an optional embodiment provided by the present invention, the first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix are input into the current transaction risk identification model to obtain the transaction risk value of the current transaction, including:
[0142] Inputting the first feature data matrix, the second feature data matrix and the third feature data matrix into the current transaction risk identification model, and obtaining data convolution features corresponding to the first feature data matrix, the second feature data matrix and the third feature data matrix respectively through different convolution layers in the current transaction risk identification model;
[0143] Input any two data convolution features into the corresponding prediction module to obtain the corresponding prediction value;
[0144] The transaction risk value of the current transaction to be traded is obtained according to the obtained prediction value.
[0145] In an optional embodiment provided by the present invention, the determination module 23 is specifically configured to:
[0146] The transaction risk value of the current transaction to be traded is calculated based on each prediction value and its corresponding prediction probability value, as well as the accuracy of each prediction module.
[0147] In an optional embodiment provided by the present invention, the training process of the current transaction risk identification model is:
[0148] Obtaining sample data for training the current transaction risk identification model and marking risk values; the sample data includes a first sample data matrix, a second sample data matrix, and a third sample data matrix;
[0149] Inputting the first sample data matrix, the second sample data matrix, and the third sample data matrix into the current transaction risk identification model to obtain a first prediction value, a second prediction value, and a third prediction value;
[0150] Calculating a final prediction value according to the first prediction value, the second prediction value and the third prediction value;
[0151] The loss value is calculated by using the final predicted value and the marked risk value. When the loss value is less than the target loss value, the training of the current transaction risk identification model is completed.
[0152] For the specific definition of the device, please refer to the definition of the transaction risk assessment method based on artificial intelligence above, which will not be repeated here. Each module in the above device can be implemented in whole or in part by software, hardware and their combination. Each of the above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.
[0153] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a transaction risk assessment method based on artificial intelligence is implemented.
[0154] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program:
[0155] Acquire the transaction data information of the current transaction to be traded, wherein the transaction data information includes the current buyer data information, the current seller data information and the current transaction commodity data information;
[0156] Acquire the buyer's historical data information corresponding to the current buyer's data information, and the seller's historical data information corresponding to the current seller's data information;
[0157] Extracting buyer data statistical features and buyer transaction risk features based on the current buyer data information and the buyer historical data information; and extracting seller data statistical features and seller transaction risk features based on the current seller data information and the seller historical data information;
[0158] Determine a first characteristic data matrix according to the buyer's data statistical characteristics and the buyer's transaction risk characteristics, determine a second characteristic data matrix according to the seller's data statistical characteristics and the seller's transaction risk characteristics, and determine a third characteristic data matrix according to the current transaction commodity data information;
[0159] Determining the transaction risk value of the current transaction to be traded according to the first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix;
[0160] When the risk value of the transaction currently to be traded is greater than the preset risk value, the transaction currently to be traded is prohibited.
[0161] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:
[0162] Acquire the transaction data information of the current transaction to be traded, wherein the transaction data information includes the current buyer data information, the current seller data information and the current transaction commodity data information;
[0163] Acquire the buyer's historical data information corresponding to the current buyer's data information, and the seller's historical data information corresponding to the current seller's data information;
[0164] Extracting buyer data statistical features and buyer transaction risk features based on the current buyer data information and the buyer historical data information; and extracting seller data statistical features and seller transaction risk features based on the current seller data information and the seller historical data information;
[0165] Determine a first characteristic data matrix according to the buyer's data statistical characteristics and the buyer's transaction risk characteristics, determine a second characteristic data matrix according to the seller's data statistical characteristics and the seller's transaction risk characteristics, and determine a third characteristic data matrix according to the current transaction commodity data information;
[0166] Determining the transaction risk value of the current transaction to be traded according to the first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix;
[0167] When the risk value of the transaction currently to be traded is greater than the preset risk value, the transaction currently to be traded is prohibited.
[0168] In one embodiment, a computer program product is provided, the computer program product comprising a computer program, the computer program being executed by a processor to implement the following steps:
[0169] Acquire the transaction data information of the current transaction to be traded, wherein the transaction data information includes the current buyer data information, the current seller data information and the current transaction commodity data information;
[0170] Acquire the buyer's historical data information corresponding to the current buyer's data information, and the seller's historical data information corresponding to the current seller's data information;
[0171] Extracting buyer data statistical features and buyer transaction risk features based on the current buyer data information and the buyer historical data information; and extracting seller data statistical features and seller transaction risk features based on the current seller data information and the seller historical data information;
[0172] Determine a first characteristic data matrix according to the buyer's data statistical characteristics and the buyer's transaction risk characteristics, determine a second characteristic data matrix according to the seller's data statistical characteristics and the seller's transaction risk characteristics, and determine a third characteristic data matrix according to the current transaction commodity data information;
[0173] Determining the transaction risk value of the current transaction to be traded according to the first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix;
[0174] When the risk value of the transaction currently to be traded is greater than the preset risk value, the transaction currently to be traded is prohibited.
[0175] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0176] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.
[0177] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.
Claims
1. A transaction risk assessment method based on artificial intelligence, characterized in that: The method comprises: Acquire the transaction data information of the current transaction to be traded, wherein the transaction data information includes the current buyer data information, the current seller data information and the current transaction commodity data information; Acquire the buyer's historical data information corresponding to the current buyer's data information, and the seller's historical data information corresponding to the current seller's data information; Extracting buyer data statistical features and buyer transaction risk features based on the current buyer data information and the buyer historical data information; and extracting seller data statistical features and seller transaction risk features based on the current seller data information and the seller historical data information; Determine a first characteristic data matrix according to the buyer's data statistical characteristics and the buyer's transaction risk characteristics, determine a second characteristic data matrix according to the seller's data statistical characteristics and the seller's transaction risk characteristics, and determine a third characteristic data matrix according to the current transaction commodity data information; Determining the transaction risk value of the current transaction to be traded according to the first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix; When the risk value of the transaction currently to be traded is greater than the preset risk value, the transaction currently to be traded is prohibited.
2. The method according to claim 1, characterized in that The extracting of buyer data statistical features and buyer transaction risk features according to the current buyer data information and the buyer historical data information includes: Extracting data from the current buyer data information and the buyer historical data information respectively to obtain buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount; Determine the buyer data statistical characteristics based on the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount; The keyword matching result is obtained by performing keyword matching on the buyer transaction type, buyer device basic information, and buyer device usage information of the current buyer data information through the keywords in the first mapping table, wherein the first mapping table stores a plurality of keywords and their corresponding risk values; The buyer's transaction risk characteristics are determined based on the keyword matching results and their corresponding risk values.
3. The method according to claim 2, characterized in that The buyer data statistical features are determined based on the obtained buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount, including: Matching the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount respectively corresponding to the current buyer data information and the buyer historical data information to obtain a data item matching result, and obtaining first weight values respectively corresponding to the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount according to the data item matching result; Matching the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount of the current buyer data information through the weight information table to determine second weight values corresponding to the buyer transaction type, buyer device basic information, buyer device usage information, and buyer transaction amount, respectively; The buyer data statistical characteristics are determined according to the buyer transaction type, buyer device basic information, buyer device usage information, buyer transaction amount and their corresponding first weight values and second weight values respectively.
4. The method according to claim 1, characterized in that: The extracting of seller data statistical features and seller transaction risk features according to the current seller data information and the seller historical data information includes: Extracting data from the current seller data information and the seller historical data information respectively to obtain seller transaction records, seller payment records, and seller transaction methods; Determine the seller data statistical features according to the seller transaction records, seller payment records and seller transaction methods respectively corresponding to the current seller data information and the seller historical data information; Convert the seller transaction record, seller payment record and seller transaction method of the current seller data information into a first feature vector, and convert the seller transaction record, seller payment record and seller transaction method of the seller historical data information into a second feature vector; Inputting the first feature vector and the second feature vector into a seller transaction risk identification model to obtain risk weight values corresponding to the seller transaction record, seller payment record, and seller transaction method of the current seller data information; The seller data statistical characteristics are determined based on the seller transaction records, seller payment records and seller transaction methods of the current seller data information and their corresponding risk weight values.
5. The method according to claim 4, characterized in that The inputting the first feature vector and the second feature vector into the seller transaction risk identification model to obtain the risk weight values corresponding to the seller transaction record, the seller payment record and the seller transaction method of the current seller data information respectively include: Inputting the first feature vector and the second feature vector into a seller's transaction risk identification model to obtain a transaction risk prediction value; According to the importance of corresponding data items in the first feature vector and the second feature vector to obtaining the transaction risk prediction value, the risk weight values corresponding to the seller transaction record, the seller payment record and the seller transaction method of the current seller data information are determined.
6. The method according to any one of claims 1 to 5, characterized in that: The step of determining the transaction risk value of the current transaction to be traded according to the first characteristic data matrix, the second characteristic data matrix, and the third characteristic data matrix includes: The first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix are input into the current transaction risk identification model to obtain the transaction risk value of the current transaction.
7. The method according to claim 6, characterized in that The step of inputting the first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix into the current transaction risk identification model to obtain the transaction risk value of the current transaction to be traded comprises: Inputting the first feature data matrix, the second feature data matrix and the third feature data matrix into the current transaction risk identification model, and obtaining data convolution features corresponding to the first feature data matrix, the second feature data matrix and the third feature data matrix respectively through different convolution layers in the current transaction risk identification model; Input any two data convolution features into the corresponding prediction module to obtain the corresponding prediction value; The transaction risk value of the current transaction to be traded is obtained according to the obtained prediction value.
8. The method according to claim 7, characterized in that The step of obtaining the transaction risk value of the current transaction to be traded according to the obtained prediction value includes: The transaction risk value of the current transaction to be traded is calculated based on each prediction value and its corresponding prediction probability value, as well as the accuracy of each prediction module.
9. The method according to claim 7, characterized in that: The training process of the current transaction risk identification model is as follows: Obtaining sample data for training the current transaction risk identification model and marking risk values; The sample data includes a first sample data matrix, a second sample data matrix, and a third sample data matrix; Inputting the first sample data matrix, the second sample data matrix, and the third sample data matrix into the current transaction risk identification model to obtain a first prediction value, a second prediction value, and a third prediction value; Calculating a final prediction value according to the first prediction value, the second prediction value and the third prediction value; The loss value is calculated by using the final predicted value and the marked risk value. When the loss value is less than the target loss value, the training of the current transaction risk identification model is completed.
10. A transaction risk assessment device based on artificial intelligence, characterized in that: The device comprises: An acquisition module is used to acquire the transaction data information of the current transaction to be traded, wherein the transaction data information includes the current buyer data information, the current seller data information and the current transaction commodity data information; The acquisition module is further used to acquire the buyer's historical data information corresponding to the current buyer's data information, and the seller's historical data information corresponding to the current seller's data information; An extraction module, configured to extract buyer data statistical features and buyer transaction risk features based on the current buyer data information and the buyer historical data information; and to extract seller data statistical features and seller transaction risk features based on the current seller data information and the seller historical data information; A determination module, configured to determine a first characteristic data matrix according to the buyer's data statistical characteristics and the buyer's transaction risk characteristics, determine a second characteristic data matrix according to the seller's data statistical characteristics and the seller's transaction risk characteristics, and determine a third characteristic data matrix according to the current transaction commodity data information; The determination module is further used to determine the transaction risk value of the current pending transaction based on the first characteristic data matrix, the second characteristic data matrix and the third characteristic data matrix; The prohibition module is used to prohibit the transaction currently to be traded when the risk value of the transaction currently to be traded is greater than the preset risk value.