Transaction data information category determination method, device, and equipment, and storage medium

Abstract

The application provides a transaction data information category determination method and device, equipment and a storage medium, and relates to the field of big data. The method comprises the following steps: obtaining transaction data information to be determined, inputting the transaction data information into a trained category recognition model to obtain a first result; inputting the transaction data information into a preset rule to obtain a second result; determining the category of the transaction data information according to the first result and the second result; and wherein the trained category recognition model is trained based on at least one first transaction data information, at least one second transaction data information and a preset rule. By using the technical solution, the dependence on the rules prepared by human beings in advance can be reduced, the change of the business environment can be better adapted, the manual labor required for adjusting the rule model can be reduced, and the transaction data information of money laundering can be identified.

Description

Technical Field

[0001] This application relates to the field of big data, and in particular to a method, apparatus, device, and storage medium for determining the category of transaction data information. Background Technology

[0002] Currently, existing anti-money laundering models are primarily built using rule-based identification methods. Traditional rule engines rely on business rules summarized from human experience, requiring business experts to define explanatory variables. With the development of internet finance, the types of businesses have become more diverse, the proportion of online business has increased, and the volume of business has grown. These changes in the business environment have rendered traditional rule engines unable to meet current needs.

[0003] Therefore, there is an urgent need for a method to classify transaction data information that can rely less on pre-defined rules, better adapt to changes in the business environment, reduce the manual labor required to adjust rule models, and identify money laundering transaction data information. Summary of the Invention

[0004] This application provides a method, apparatus, device, and storage medium for determining the category of transaction data information, which can rely less on pre-defined rules, better adapt to changes in the business environment, reduce the manual labor required to adjust the rule model, and identify money laundering transaction data information.

[0005] Firstly, this application provides a method for determining the category of transaction data information, the method comprising:

[0006] Obtain transaction data information for the category to be determined;

[0007] The transaction data information is input into the trained category recognition model to obtain a first result;

[0008] The transaction data information is input into a preset rule to obtain a second result;

[0009] Based on the first result and the second result, the category of the transaction data information is determined; wherein, the trained category recognition model is trained based on at least one first transaction data information, at least one second transaction data information, and preset rules; wherein, the first transaction data information represents transaction data information with labels; and the second transaction data information represents transaction data information without labels.

[0010] In one example, determining the category of the transaction data information based on the first result and the second result includes:

[0011] If the difference between the first result and the second result is greater than a threshold, then cluster analysis is performed on the first result and the second result to obtain the clustering result;

[0012] The preset rules are updated based on the clustering results.

[0013] In one example, updating the preset rule based on the clustering results includes:

[0014] Based on the clustering results, a third data feature is determined;

[0015] If the third data feature is not in the preset rule, then the third data feature is added to the preset rule.

[0016] In one example, the trained category recognition model is trained based on at least one first transaction data piece, at least one second transaction data piece, and preset rules, including:

[0017] Based on preset rules, the label of the second transaction data information is determined, and the third transaction data information is obtained; wherein, the third transaction data information represents transaction data information with a label;

[0018] The first transaction data information and the third transaction data information are input into the category recognition model to train the category recognition model and obtain the trained category recognition model; wherein, the first transaction data information represents transaction data information with labels; and the second transaction data information represents transaction data information without labels.

[0019] Secondly, this application provides a method for training a category recognition model, including:

[0020] Obtain at least one first transaction data piece and at least one second transaction data piece; wherein, the first transaction data piece represents transaction data with a tag; and the second transaction data piece represents transaction data without a tag;

[0021] According to preset rules, the label of the second transaction data information is determined, and the third transaction data information is obtained; wherein, the third transaction data information represents transaction data information with a label;

[0022] The first transaction data information and the third transaction data information are input into the category recognition model to train the category recognition model and obtain the trained category recognition model; wherein, the trained category recognition model is used to determine the category of the transaction data information.

[0023] In one example, inputting the first transaction data information and the third transaction data information into a category recognition model to train the category recognition model includes:

[0024] Extract the first data feature from the first transaction data information; wherein, the first data feature characterizes the object, transaction amount, and transaction frequency of the first transaction data information;

[0025] Extract the second data feature from the third transaction data information; wherein, the second data feature characterizes the object, transaction amount, and transaction frequency of the third transaction data information;

[0026] The first data feature and the second data feature are input into the category recognition model to train the category recognition model.

[0027] In one example, the second data feature for extracting the third transaction data information includes:

[0028] Based on the preset rules, the second data features of the third transaction data information are extracted.

[0029] In one example, after obtaining at least one piece of first transaction data and at least one piece of second transaction data, the method further includes:

[0030] The at least one first transaction data information and the at least one second transaction data information are processed to obtain the processed transaction data information.

[0031] In one example, inputting the first data feature and the second data feature into the category recognition model to train the category recognition model includes:

[0032] Based on the occlusion task in the category recognition model, the first data features are processed to obtain a first loss function;

[0033] Based on the statement relationship task in the category recognition model, the second data features are processed to obtain the second loss function;

[0034] Based on the first loss function and the second loss function, the loss function of the category recognition model is obtained;

[0035] The category recognition model is trained based on the loss function of the category recognition model.

[0036] In one example, the processing of the second data features based on the statement relationship task in the category recognition model to obtain the second loss function includes:

[0037] Based on the statement relationship task in the category recognition model, the second data features are processed to obtain the predicted value;

[0038] The second loss function is determined based on the distance between the predicted value and the true value; wherein the true value represents the label of the second transaction data information.

[0039] In one example, the second data feature for extracting the third transaction data information includes:

[0040] Based on the preset rules, the second data features of the third transaction data information are extracted.

[0041] In one example, the first data feature for extracting the first transaction data information includes:

[0042] Based on the convolutional layer in the category recognition model, the feature vector of the first transaction data information is extracted, and the feature vector is used as the first data feature.

[0043] Thirdly, this application provides a device for determining the category of transaction data information, the device comprising:

[0044] The first acquisition unit is used to acquire transaction data information of the category to be determined;

[0045] The first determining unit is used to input the transaction data information into the trained category recognition model to obtain a first result;

[0046] The second determining unit is used to input the transaction data information into a preset rule to obtain a second result;

[0047] The third determining unit is used to determine the category of the transaction data information based on the first result and the second result; wherein the trained category recognition model is trained based on at least one first transaction data information, at least one second transaction data information and preset rules; wherein the first transaction data information represents transaction data information with labels; and the second transaction data information represents transaction data information without labels.

[0048] In one example, the third determining unit includes:

[0049] The clustering module is used to perform clustering analysis on the first result and the second result if the difference between the first result and the second result is greater than a threshold, and obtain the clustering result.

[0050] An update module is used to update the preset rules based on the clustering results.

[0051] In one example, the update module includes:

[0052] A determination submodule is used to determine a third data feature based on the clustering results;

[0053] A submodule is added to the preset rule if the third data feature is not in the preset rule.

[0054] In one example, the device includes:

[0055] The fourth determining unit is used to determine the tag of the second transaction data information based on preset rules, and obtain the third transaction data information; wherein the third transaction data information represents transaction data information with tags;

[0056] The first training unit is used to input the first transaction data information and the third transaction data information into the category recognition model to train the category recognition model and obtain the trained category recognition model; wherein, the first transaction data information represents transaction data information with labels; and the second transaction data information represents transaction data information without labels.

[0057] Fourthly, this application provides a training apparatus for a category recognition model, comprising:

[0058] The second acquisition unit is used to acquire at least one first transaction data information and at least one second transaction data information; wherein, the first transaction data information represents transaction data information with a tag; and the second transaction data information represents transaction data information without a tag.

[0059] The fifth determining unit is used to determine the tag of the second transaction data information according to a preset rule, and obtain the third transaction data information; wherein the third transaction data information represents transaction data information with a tag;

[0060] The second training unit is used to input the first transaction data information and the third transaction data information into the category recognition model to train the category recognition model and obtain the trained category recognition model; wherein, the trained category recognition model is used to determine the category of the transaction data information.

[0061] In one example, the second training unit includes:

[0062] The first extraction module is used to extract the first data feature of the first transaction data information; wherein, the first data feature represents the object, transaction amount and transaction frequency of the first transaction data information;

[0063] The second extraction module is used to extract the second data features of the third transaction data information; wherein, the second data features characterize the object, transaction amount, and transaction frequency of the third transaction data information;

[0064] The training module is used to input the first data feature and the second data feature into the category recognition model to train the category recognition model.

[0065] In one example, the training module includes:

[0066] The first determining submodule is used to process the first data features based on the occlusion task in the category recognition model to obtain a first loss function;

[0067] The second determination submodule is used to process the second data features based on the statement relationship task in the category recognition model to obtain the second loss function;

[0068] The third determining submodule is used to obtain the loss function of the category recognition model based on the first loss function and the second loss function;

[0069] The training submodule is used to train the category recognition model based on the loss function of the category recognition model.

[0070] In one example, the second determined submodule is specifically used for:

[0071] Based on the statement relationship task in the category recognition model, the second data features are processed to obtain the predicted value;

[0072] The second loss function is determined based on the distance between the predicted value and the true value; wherein the true value represents the label of the second transaction data information.

[0073] In one example, the second extraction module is specifically used for:

[0074] Based on the preset rules, the second data features of the third transaction data information are extracted.

[0075] In one example, the first extraction module is specifically used for:

[0076] Based on the convolutional layer in the category recognition model, the feature vector of the first transaction data information is extracted, and the feature vector is used as the first data feature.

[0077] In one example, the device also includes:

[0078] The processing unit is used to process the at least one first transaction data information and the at least one second transaction data information to obtain processed transaction data information.

[0079] Fifthly, this application provides an electronic device, including: a processor, and a memory communicatively connected to the processor;

[0080] The memory stores computer-executed instructions;

[0081] The processor executes computer execution instructions stored in the memory to implement the method as described in the first or second aspect.

[0082] In a sixth aspect, this application provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, are used to implement the method described in the first or second aspect.

[0083] This application provides a method, apparatus, device, and storage medium for determining the category of transaction data information. The method involves acquiring transaction data information of a desired category, inputting the transaction data information into a trained category recognition model to obtain a first result, inputting the transaction data information into preset rules to obtain a second result, and determining the category of the transaction data information based on the first and second results. The trained category recognition model is trained based on at least one first transaction data piece, at least one second transaction data piece, and preset rules. The first transaction data information represents transaction data information with labels, and the second transaction data information represents transaction data information without labels. This technical solution reduces reliance on manually pre-defined rules, better adapts to changes in the business environment, reduces the manual labor required to adjust the rule model, and can identify money laundering transaction data information. Attached Figure Description

[0084] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0085] Figure 1 This is a flowchart illustrating a method for determining the category of transaction data information according to Embodiment 1 of this application;

[0086] Figure 2 This is a flowchart illustrating a training method for a category recognition model according to Embodiment 2 of this application;

[0087] Figure 3 This is a schematic diagram of a device for determining the category of transaction data information according to Embodiment 3 of this application;

[0088] Figure 4This is a schematic diagram of a device for determining the category of transaction data information according to Embodiment 4 of this application;

[0089] Figure 5 This is a schematic diagram of the structure of a training device for a category recognition model according to Embodiment 5 of this application;

[0090] Figure 6 This is a block diagram illustrating an electronic device according to an exemplary embodiment.

[0091] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concepts of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0092] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0093] It should be noted that the user information (including but not limited to user device information, user personal information and transaction data information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with relevant laws, regulations and standards, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0094] It should be noted that the method and apparatus for determining the category of transaction data information provided in this application can be used in the field of big data, or in any field other than big data. The application field of the training method and apparatus for the category recognition model provided in this application is not limited.

[0095] Currently, in the field of financial risk control, money laundering models using fund flow network data mining can be mainly divided into three types: the first is manual identification, which requires salespersons to review each transaction; the second is rule engine identification, which involves experts setting anti-money laundering rules, capturing violators, and quantifying the severity of the violations; and the third is intelligent engine identification, which combines machine learning, knowledge graphs, and other technologies for comprehensive judgment.

[0096] While existing technologies offer solutions combining machine learning models and expert rules, they often result in high false positive rates in real-world anti-money laundering scenarios, still requiring significant manual review. This is primarily due to several issues: large data scale, sparse labels, and uninterpretable models.

[0097] This application provides a method for determining the category of transaction data information, which aims to solve the above-mentioned technical problems in the prior art.

[0098] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0099] Figure 1 This is a flowchart illustrating a method for determining the category of transaction data information according to Embodiment 1 of this application.

[0100] Example 1 includes the following steps:

[0101] S101. Obtain transaction data information for the category to be determined.

[0102] In this embodiment, the transaction data information whose category needs to be determined can be transaction data information retrieved from the system, and the category of this transaction data information is unknown.

[0103] S102. Input the transaction data information into the trained category recognition model to obtain the first result.

[0104] In this embodiment, the trained category recognition model can determine the category of the transaction data information to be classified and obtain a first result, wherein the first result can be a score, which can be a value between 0 and 1. If the score is 0.9, it indicates that the transaction data information is money laundering transaction data information; if the score is 0.1, it indicates that the transaction data information is normal transaction data information.

[0105] In one example, the trained category recognition model is trained based on at least one first transaction data piece, at least one second transaction data piece, and preset rules, including:

[0106] Based on preset rules, the label of the second transaction data information is determined, and the third transaction data information is obtained; wherein, the third transaction data information represents transaction data information with a label;

[0107] The first transaction data information and the third transaction data information are input into the category recognition model to train the category recognition model and obtain the trained category recognition model; wherein, the first transaction data information represents transaction data information with labels; and the second transaction data information represents transaction data information without labels.

[0108] In this embodiment, both the first transaction data information and the second transaction data information are historical transaction data information retrieved from the system. The first transaction data information is tagged transaction data information, where the tag represents the category of the first transaction data information; for example, the tag can be money laundering transaction data information or non-money laundering transaction data information. Furthermore, the second transaction data information represents transaction data information without tags. The second transaction data information can be transaction data information that has not been tagged, which may be transaction data information without tags or transaction data information whose tags have not been identified.

[0109] In this embodiment, the preset rules are discrimination conditions or combinations of discrimination conditions given by professionals, or they can refer to a discriminator that gives a unique result for a given input. The discriminator can be an existing mature model or a combination of models.

[0110] In this embodiment, a label is determined for the second transaction data information according to preset rules, thus obtaining the third transaction data information with the label. Specifically, using preset rules as constraints, an evaluation value for the second transaction data information is calculated. Based on the evaluation value and the black-and-white sample discrimination threshold, the second transaction data information is labeled. For example, the money laundering suspicion of the second transaction data information is scored using preset rules. If the score is 0.9, the label for the second transaction data information is money laundering transaction data information; if the score is 0.1, the label for the second transaction data information is ordinary transaction data information.

[0111] In this embodiment, the category recognition model can be a graph neural network model, a machine learning model, or a deep learning model. The category recognition model can be trained using the first transaction data information and its corresponding label, and the third transaction data information and its corresponding label, enabling the trained category recognition model to accurately output the category of the transaction data information. For example, the trained category recognition model can determine whether the transaction data information is money laundering transaction data information.

[0112] S103. Input the transaction data information into the preset rules to obtain the second result.

[0113] In this embodiment, the transaction data information to be categorized is also input into preset rules. These rules score the transaction data information to be categorized, yielding a second result, which can be a score. The score can be a value between 0 and 1. If the score is 0.9, it indicates that the transaction data information is money laundering; if the score is 0.1, it indicates that the transaction data information is normal.

[0114] S104. Determine the category of transaction data information based on the first result and the second result.

[0115] In this embodiment, since the first result and the second result are for the same transaction data information of the same category to be determined, the first result and the second result should have the same score, and the category of the transaction data information is determined based on the score.

[0116] In one example, based on the first and second results, the categories of transaction data information are determined, including:

[0117] If the difference between the first result and the second result is greater than the threshold, then cluster analysis is performed on the first result and the second result to obtain the cluster result;

[0118] Update the preset rules based on the clustering results.

[0119] In this embodiment, the threshold can be set by the user and can be 0.5. If the difference between the first result and the second result is greater than the threshold, it indicates that the preset rules may be unreasonable, and the preset rules need to be adjusted. Specifically, if the difference between the first result and the second result is greater than the threshold, cluster analysis is performed on the first result and the second result. Specifically, a clustering algorithm similar to k-means can be used to perform cluster analysis on the first result and the second result, and the preset rules are updated based on the final clustering results. Updating the preset rules can be done by adding new rules or modifying the preset rules.

[0120] In one example, based on the clustering results, the preset rules are updated, including:

[0121] Based on the clustering results, the third data feature is determined;

[0122] If the third data feature is not in the preset rules, then add the third data feature to the preset rules.

[0123] In this embodiment, a third data feature can be obtained through clustering results. The third data feature is then matched with a preset rule using the SVM method. If a match is found, the weight of the preset rule is updated according to the influence of the feature on the result. If no match is found, the feature is added to the preset rule as a new rule.

[0124] This application provides a method for determining the category of transaction data information. The method involves acquiring transaction data information of a desired category, inputting the transaction data information into a trained category recognition model to obtain a first result, inputting the transaction data information into preset rules to obtain a second result, and then determining the category of the transaction data information based on the first and second results. This technical solution effectively mines money laundering transaction data information, and the implementation process is relatively simple.

[0125] Figure 2 This is a flowchart illustrating a training method for a category recognition model according to Embodiment 2 of this application. Embodiment 2 includes the following steps:

[0126] S201. Obtain at least one first transaction data information and at least one second transaction data information; wherein, the first transaction data information represents transaction data information with a tag; and the second transaction data information represents transaction data information without a tag.

[0127] In one example, after obtaining at least one piece of first transaction data and at least one piece of second transaction data, the method further includes:

[0128] Data processing is performed on at least one first transaction data piece and at least one second transaction data piece to obtain processed transaction data.

[0129] In this embodiment, data processing may involve performing anomaly processing on at least one first transaction data piece and at least one second transaction data piece. The anomaly data may be null values, transaction data that is above a threshold, or transaction data that is below a threshold. In this embodiment, after processing the first and second transaction data pieces, processed transaction data is obtained, and further processing is then performed on the processed transaction data.

[0130] S202. According to the preset rules, determine the label of the second transaction data information and obtain the third transaction data information; wherein, the third transaction data information represents the transaction data information with the label.

[0131] For example, this step can refer to step S102 above, and will not be repeated here.

[0132] S203. Extract the first data feature of the first transaction data information; wherein the first data feature represents the object, transaction amount and transaction frequency of the first transaction data information.

[0133] In this embodiment, graph structure data information can be constructed from the first transaction data information, and data features of the graph structure data information can be extracted and used as the first data feature. The data features of the graph structure data information can include the in-degree and out-degree of nodes, the number of neighboring nodes, etc. The first data feature characterizes the object, transaction amount, and transaction frequency of the first transaction data information.

[0134] S204. Extract the second data features of the third transaction data information; wherein, the second data features characterize the object, transaction amount and transaction frequency of the third transaction data information.

[0135] In one example, the second data feature extracted from the third transaction data information includes:

[0136] Based on preset rules, the second data features of the third transaction data information are extracted.

[0137] In this embodiment, the preset rules are the same as those in step S102. The second data features of the third transaction data information are determined by the preset rules. Specifically, the second data features can be determined by content matching or other methods.

[0138] S205. Input the first data feature and the second data feature into the category recognition model to train the category recognition model.

[0139] In this embodiment, after the first data feature and the second data feature are input into the category recognition model, the parameter information in the category recognition model is adjusted, thereby training the category recognition model.

[0140] In one example, inputting the first data feature and the second data feature into the category recognition model to train the category recognition model includes:

[0141] Based on the occlusion task in the category recognition model, the first data features are processed to obtain a first loss function;

[0142] Based on the statement relationship task in the category recognition model, the second data features are processed to obtain the second loss function;

[0143] Based on the first loss function and the second loss function, the loss function of the category recognition model is obtained;

[0144] The category recognition model is trained based on the loss function of the category recognition model.

[0145] In this embodiment, the occlusion task involves processing a preset portion of the second data feature, where the preset portion can be 15%. Specifically, the processing method is to occlude 80% of the second data feature; replace 10% of the processed second data feature; and maintain the original method for the remaining 10% of the processed second data feature.

[0146] In this embodiment, the statement relationship task refers to the task of identifying the relationships between statements. There are various types of statement relationships in the statement relationship task, and the relationships between statements are different in each type. In this embodiment, the first loss function, the second loss function, or the loss function of the category recognition model can be the Softmax function.

[0147] In this embodiment, the squares of the first loss function and the second loss function in the normalization result are calculated, and the squares of the first loss function and the second loss function are summed to obtain the sum value. Based on the sum value, the loss function of the category recognition model is determined, and the category recognition model is trained based on the loss function of the category recognition model.

[0148] In one example, based on the statement relationship task in the category recognition model, the second data features are processed to obtain the second loss function, which includes:

[0149] Based on the sentence relationship task in the category recognition model, the second data features are processed to obtain the predicted value;

[0150] A second loss function is determined based on the distance between the predicted value and the true value; where the true value represents the label of the second transaction data information.

[0151] In this embodiment, the predicted value represents the label of the third transaction data information determined by the category recognition model, and the true value represents the label of the second transaction data information. The second loss function is determined by the difference between the two, where the difference can be determined by the distance value, which represents the similarity between the two.

[0152] In one example, the first data features for extracting the first transaction data information include:

[0153] Based on the convolutional layer in the category recognition model, the feature vector of the first transaction data information is extracted, and the feature vector is used as the first data feature.

[0154] In this embodiment, after obtaining the first transaction data information, the convolutional layer extracts the feature vector of the first transaction data information. The feature vector can be a position feature vector, a phrase feature vector, or a word feature vector.

[0155] This application provides a training method for a category recognition model. The method involves determining the label of second transaction data information according to preset rules, obtaining third transaction data information, extracting first data features from the first transaction data information, extracting second data features from the third transaction data information, and inputting the first and second data features into the category recognition model to train the model. This technical solution can address the problems of sparse labels and class imbalance in anti-money laundering modeling scenarios.

[0156] Figure 3 This is a schematic diagram of a device for determining the category of transaction data information according to Embodiment 3 of this application. Specifically, the device 30 in Embodiment 3 includes:

[0157] The first acquisition unit 301 is used to acquire transaction data information of the category to be determined.

[0158] The first determining unit 302 is used to input the transaction data information into the trained category recognition model to obtain a first result.

[0159] The second determining unit 303 is used to input the transaction data information into a preset rule to obtain a second result.

[0160] The third determining unit 304 is used to determine the category of the transaction data information based on the first result and the second result; wherein the trained category recognition model is trained based on at least one first transaction data information, at least one second transaction data information and preset rules; wherein the first transaction data information represents transaction data information with labels; and the second transaction data information represents transaction data information without labels.

[0161] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described device can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0162] Figure 4 This is a schematic diagram of a device for determining the category of transaction data information according to Embodiment 4 of this application. Specifically, the device 40 in Embodiment 4 includes:

[0163] The first acquisition unit 401 is used to acquire transaction data information of the category to be determined.

[0164] The first determining unit 402 is used to input the transaction data information into the trained category recognition model to obtain a first result.

[0165] The second determining unit 403 is used to input the transaction data information into a preset rule to obtain a second result.

[0166] The third determining unit 404 is used to determine the category of the transaction data information based on the first result and the second result; wherein the trained category recognition model is trained based on at least one first transaction data information, at least one second transaction data information and preset rules; wherein the first transaction data information represents transaction data information with labels; and the second transaction data information represents transaction data information without labels.

[0167] In one example, the third determining unit 404 includes:

[0168] Clustering module 4041 is used to perform clustering analysis on the first result and the second result if the difference between the first result and the second result is greater than a threshold, and obtain clustering results;

[0169] The update module 4042 is used to update the preset rules based on the clustering results.

[0170] In one example, updating module 4042 includes:

[0171] The determination submodule 40421 is used to determine the third data feature based on the clustering results;

[0172] Add submodule 40422, which is used to add the third data feature to the preset rule if the third data feature is not in the preset rule.

[0173] In one example, the device 40 includes:

[0174] The fourth determining unit 405 is used to determine the tag of the second transaction data information based on preset rules and obtain the third transaction data information; wherein the third transaction data information represents transaction data information with tags.

[0175] The first training unit 406 is used to input the first transaction data information and the third transaction data information into the category recognition model to train the category recognition model and obtain the trained category recognition model; wherein, the first transaction data information represents transaction data information with labels; and the second transaction data information represents transaction data information without labels.

[0176] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described device can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0177] Figure 5This is a schematic diagram of a training device for a category recognition model according to Embodiment 5 of this application. Specifically, the device 50 in Embodiment 5 includes:

[0178] The second acquisition unit 501 is used to acquire at least one first transaction data information and at least one second transaction data information; wherein, the first transaction data information represents transaction data information with a tag; and the second transaction data information represents transaction data information without a tag.

[0179] The fifth determining unit 502 is used to determine the label of the second transaction data information according to the preset rules and obtain the third transaction data information; wherein the third transaction data information represents the transaction data information with the label.

[0180] The second training unit 503 is used to input the first transaction data information and the third transaction data information into the category recognition model to train the category recognition model and obtain the trained category recognition model; wherein, the trained category recognition model is used to determine the category of the transaction data information.

[0181] In one example, the second training unit 503 includes:

[0182] The first extraction module 5031 is used to extract the first data features of the first transaction data information; wherein, the first data features characterize the object, transaction amount and transaction frequency of the first transaction data information.

[0183] The second extraction module 5032 is used to extract the second data features of the third transaction data information; wherein, the second data features characterize the object, transaction amount and transaction frequency of the third transaction data information.

[0184] The training module 5033 is used to input the first data feature and the second data feature into the category recognition model to train the category recognition model.

[0185] In one example, the second extraction module 5032 is also used to extract the second data features of the third transaction data information based on preset rules.

[0186] In one example, training module 5033 includes:

[0187] The first determining submodule 50331 is used to process the first data features based on the occlusion task in the category recognition model to obtain a first loss function.

[0188] The second determining submodule 50332 is used to process the second data features based on the statement relationship task in the category recognition model to obtain the second loss function.

[0189] The third determining submodule 50333 is used to obtain the loss function of the category recognition model based on the first loss function and the second loss function.

[0190] Training submodule 50334 is used to train the category recognition model based on the loss function of the category recognition model.

[0191] In one example, the second determining submodule 50332 is specifically used for:

[0192] Based on the statement relationship task in the category recognition model, the second data features are processed to obtain the predicted value;

[0193] The second loss function is determined based on the distance between the predicted value and the true value; wherein the true value represents the label of the second transaction data information.

[0194] In one example, the first extraction module 5031 is specifically used for:

[0195] Based on the convolutional layer in the category recognition model, the feature vector of the first transaction data information is extracted, and the feature vector is used as the first data feature.

[0196] In one example, the device 50 also includes:

[0197] The processing unit 504 is used to process the at least one first transaction data information and the at least one second transaction data information to obtain processed transaction data information.

[0198] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described device can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0199] Figure 6 This is a block diagram illustrating an electronic device according to an exemplary embodiment. The device may be a mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, medical device, fitness device, personal digital assistant, etc.

[0200] Electronic device 600 may include one or more of the following components: processing component 602, memory 604, power supply component 606, multimedia component 608, audio component 610, input / output (I / O) interface 612, sensor component 614, and communication component 616.

[0201] Processing component 602 typically controls the overall operation of electronic device 600, such as operations associated with display, telephone calls, data communication, camera operation, and recording operations. Processing component 602 may include one or more processors 620 to execute instructions to perform all or part of the steps of the methods described above. Furthermore, processing component 602 may include one or more modules to facilitate interaction between processing component 602 and other components. For example, processing component 602 may include a multimedia module to facilitate interaction between multimedia component 608 and processing component 602.

[0202] Memory 604 is configured to store various types of data to support the operation of electronic device 600. Examples of this data include instructions for any application or method operating on electronic device 600, contact data, phonebook data, messages, pictures, videos, etc. Memory 604 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0203] Power supply component 606 provides power to various components of electronic device 600. Power supply component 606 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to electronic device 600.

[0204] Multimedia component 608 includes a screen that provides an output interface between electronic device 600 and user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touchscreen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors may sense not only the boundaries of the touch or swipe action but also the duration and pressure associated with the touch or swipe operation. In some embodiments, multimedia component 608 includes a front-facing camera and / or a rear-facing camera. When electronic device 600 is in an operating mode, such as a shooting mode or video mode, the front-facing camera and / or rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0205] Audio component 610 is configured to output and / or input audio signals. For example, audio component 610 includes a microphone (MIC) configured to receive external audio signals when electronic device 600 is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 604 or transmitted via communication component 616. In some embodiments, audio component 610 also includes a speaker for outputting audio signals.

[0206] I / O interface 612 provides an interface between processing component 602 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, power buttons, and lock buttons.

[0207] Sensor assembly 614 includes one or more sensors for providing state assessments of various aspects of electronic device 600. For example, sensor assembly 614 can detect the on / off state of electronic device 600, the relative positioning of components such as the display and keypad of electronic device 600, changes in position of electronic device 600 or a component of electronic device 600, the presence or absence of user contact with electronic device 600, orientation or acceleration / deceleration of electronic device 600, and temperature changes of electronic device 600. Sensor assembly 614 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 614 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, sensor assembly 614 may also include an accelerometer, gyroscope, magnetometer, pressure sensor, or temperature sensor.

[0208] Communication component 616 is configured to facilitate wired or wireless communication between electronic device 600 and other devices. Electronic device 600 can access wireless networks based on communication standards, such as WiFi, 2G, or 3G, or combinations thereof. In one exemplary embodiment, communication component 616 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 616 also includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0209] In an exemplary embodiment, the electronic device 600 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the methods described above.

[0210] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 604 including instructions, which can be executed by a processor 620 of an electronic device 600 to perform the above-described method. For example, the non-transitory computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0211] A non-transitory computer-readable storage medium, wherein when instructions in the storage medium are executed by a processor of an electronic device, the electronic device is able to perform a method for determining the category of transaction data information of the electronic device.

[0212] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

[0213] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for determining the category of transaction data information, characterized in that, The method includes: Obtain transaction data information for the category to be determined; The transaction data information is input into the trained category recognition model to obtain a first result; The transaction data information is input into a preset rule to obtain a second result; Based on the first result and the second result, the category of the transaction data information is determined; wherein, the trained category recognition model is trained based on at least one first transaction data information, at least one second transaction data information, and preset rules; wherein, the first transaction data information represents transaction data information with labels; and the second transaction data information represents transaction data information without labels. Determining the category of the transaction data information based on the first result and the second result includes: If the difference between the first result and the second result is greater than a threshold, then cluster analysis is performed on the first result and the second result to obtain the clustering result; The preset rules are updated based on the clustering results.

2. The method according to claim 1, characterized in that, The step of updating the preset rules based on the clustering results includes: Based on the clustering results, a third data feature is determined; If the third data feature is not in the preset rule, then the third data feature is added to the preset rule.

3. The method according to claim 1, characterized in that, The trained category recognition model is trained based on at least one first transaction data piece, at least one second transaction data piece, and preset rules, including: Based on preset rules, the label of the second transaction data information is determined, and the third transaction data information is obtained; wherein, the third transaction data information represents transaction data information with a label; The first transaction data information and the third transaction data information are input into the category recognition model to train the category recognition model and obtain the trained category recognition model; wherein, the first transaction data information represents transaction data information with labels; and the second transaction data information represents transaction data information without labels.

4. The method according to claim 1, characterized in that, The training method for the category recognition model includes: Obtain at least one first transaction data piece and at least one second transaction data piece; wherein, the first transaction data piece represents transaction data with a tag; and the second transaction data piece represents transaction data without a tag; According to preset rules, the label of the second transaction data information is determined, and the third transaction data information is obtained; wherein, the third transaction data information represents transaction data information with a label; The first transaction data information and the third transaction data information are input into the category recognition model to train the category recognition model and obtain the trained category recognition model; wherein, the trained category recognition model is used to determine the category of the transaction data information.

5. The method according to claim 4, characterized in that, The step of inputting the first transaction data information and the third transaction data information into the category recognition model to train the category recognition model includes: Extract the first data feature from the first transaction data information; wherein, the first data feature characterizes the object, transaction amount, and transaction frequency of the first transaction data information; Extract the second data feature from the third transaction data information; wherein, the second data feature characterizes the object, transaction amount, and transaction frequency of the third transaction data information; The first data feature and the second data feature are input into the category recognition model to train the category recognition model.

6. The method according to claim 5, characterized in that, The step of inputting the first data feature and the second data feature into the category recognition model to train the category recognition model includes: Based on the occlusion task in the category recognition model, the first data features are processed to obtain a first loss function; Based on the statement relationship task in the category recognition model, the second data features are processed to obtain the second loss function; Based on the first loss function and the second loss function, the loss function of the category recognition model is obtained; The category recognition model is trained based on the loss function of the category recognition model.

7. The method according to claim 6, characterized in that, The task based on the statement relationship in the category recognition model processes the second data features to obtain a second loss function, including: Based on the statement relationship task in the category recognition model, the second data features are processed to obtain the predicted value; The second loss function is determined based on the distance between the predicted value and the true value; wherein the true value represents the label of the second transaction data information.

8. The method according to claim 5, characterized in that, The second data feature for extracting the third transaction data information includes: Based on the preset rules, the second data features of the third transaction data information are extracted.

9. The method according to claim 5, characterized in that, The first data feature for extracting the first transaction data information includes: Based on the convolutional layer in the category recognition model, the feature vector of the first transaction data information is extracted, and the feature vector is used as the first data feature.

10. The method according to claim 4, characterized in that, After acquiring at least one first transaction data piece and at least one second transaction data piece, the method further includes: The at least one first transaction data information and the at least one second transaction data information are processed to obtain the processed transaction data information.

11. A device for determining the category of transaction data information, characterized in that, The device includes: The first acquisition unit is used to acquire transaction data information of the category to be determined; The first determining unit is used to input the transaction data information into the category recognition model trained by the training device to obtain a first result; The second determining unit is used to input the transaction data information into a preset rule to obtain a second result; The third determining unit is used to determine the category of the transaction data information based on the first result and the second result; wherein the trained category recognition model is trained based on at least one first transaction data information, at least one second transaction data information, and preset rules; wherein the first transaction data information represents transaction data information with labels; and the second transaction data information represents transaction data information without labels. The third determining unit includes: The clustering module is used to perform clustering analysis on the first result and the second result if the difference between the first result and the second result is greater than a threshold, and obtain the clustering result. An update module is used to update the preset rules based on the clustering results.

12. The apparatus according to claim 11, characterized in that, The training device includes: The second acquisition unit is used to acquire at least one first transaction data information and at least one second transaction data information; wherein, the first transaction data information represents transaction data information with a tag; and the second transaction data information represents transaction data information without a tag. The fourth determining unit is used to determine the tag of the second transaction data information according to a preset rule, and obtain the third transaction data information; wherein the third transaction data information represents transaction data information with a tag; The training unit is used to input the first transaction data information and the third transaction data information into the category recognition model to train the category recognition model and obtain the trained category recognition model; wherein, the trained category recognition model is used to determine the category of the transaction data information.

13. An electronic device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of claims 1-10.

14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-10.

Images