Merchant operation level auditing and analyzing method, device and equipment, medium and product

Abstract

The invention provides a merchant operation level auditing analysis method and device, equipment and a storage medium, and can be applied to the technical field of artificial intelligence. The method comprises the steps of performing data extraction on pre-acquired operation data of a target merchant in a current operation period to obtain effective operation data; performing feature extraction on the effective operation data to obtain target dimension features related to merchant rating dimensions; performing merchant grade division on the target merchant according to the target dimension feature and a pre-obtained basic contrast value to obtain grade data used for representing the operation condition of the target merchant; and through a pre-trained data analysis model, according to the grade data and the target dimension features, generating an operation report with an auditing conclusion and an operation suggestion corresponding to the target merchant.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence, specifically to a merchant operation level review and analysis method, device, equipment, medium, and product. Background Technology

[0002] Enterprises or groups often have numerous merchants under their umbrella. Dynamically classifying and classifying these merchants can effectively reduce merchant inertia. Therefore, adopting modern technology to automate the review and classification of merchants is an urgent need for enterprises or groups today.

[0003] Existing technologies often suffer from unclear performance indicators, making it impossible to objectively categorize merchants' business operations and effectively guide enterprises in task allocation and targeted rewards and punishments. Even when methods exist for merchant rating, they either employ pre-defined programs to perform simple rankings based on a single calculation logic of current business data. This method only considers overall business data from a single dimension, leading to inaccurate merchant ranking and failing to fully reflect the true operational status of merchants. Alternatively, they use models to rate merchants, which only consider their general level within the broader business environment and fail to take into account their specific level among competing merchants, especially their level within their parent company or group. In other words, this rating method is too general, failing to provide enterprises or groups with a full understanding of a merchant's operational status and performance within the group, resulting in low rating accuracy and a low level of intelligence in assisting enterprise decision-making. Summary of the Invention

[0004] In view of the above problems, this application provides a merchant operation level review and analysis method, device, equipment, medium and product.

[0005] According to the first aspect of this application, a merchant operation level review and analysis method is provided, comprising: extracting data from the pre-acquired operational data of a target merchant in the current operational cycle to obtain effective operational data; extracting features from the effective operational data to obtain target dimension features related to merchant rating dimensions, wherein the merchant rating dimensions include at least one of the following: net profit of transaction items, net profit of transaction settlement, net profit of handling fees, total net profit, transaction amount, and number of transactions; classifying the target merchant into merchant levels based on the target dimension features and pre-acquired baseline reference values ​​to obtain level data characterizing the operational status of the target merchant; wherein the baseline reference values ​​are obtained by clustering preset historical operational data, wherein the historical operational data is the operational data generated by the merchant group, including the target merchant, in the previous operational cycle of the current operational cycle; and generating an operational report with review conclusions and business recommendations corresponding to the target merchant based on the level data and target dimension features using a pre-trained data analysis model.

[0006] According to an embodiment of this application, a pre-trained rating classification model uses a rating module to classify target merchants into ratings based on target dimension features and pre-acquired baseline reference values, thereby obtaining rating data to characterize the operational status of the target merchants. This includes: obtaining dimension reference values ​​corresponding to the target dimension features from the baseline reference values, where each dimension reference value represents a one-to-one correspondence with the merchant rating dimension contained in the baseline reference values; performing weighted calculations based on the dimension values ​​in the target dimension features, the dimension reference values, and the dimension weights corresponding to the target dimension features to obtain dimension weights, where each dimension weight is either a first dimension weight or a second dimension weight; summing the dimension weights corresponding to each target dimension feature to obtain merchant indicator data; and determining the rating data corresponding to the merchant indicator data in a pre-set indicator rating mapping table.

[0007] According to an embodiment of this application, when the dimension value is less than or equal to the dimension reference value, the dimension weight is the first dimension weight; wherein, calculating the first dimension weight includes: dividing the dimension value by the dimension reference value to obtain the operational reference ratio value; multiplying the operational reference ratio value by the dimension weight that corresponds one-to-one with the target dimension feature to obtain the first dimension weight; when the dimension weight is greater than the dimension reference value, the dimension weight is the second dimension weight, and the second dimension weight is the first dimension weight plus one.

[0008] According to an embodiment of this application, a basic reference value is obtained by clustering preset historical operational data through a reference module in a grading model. This includes: retrieving historical operational data related to the target merchant and extracting features from the historical operational data to obtain historical features; clustering the historical features based on a preset expected proportion of tiers to obtain merchant reference tiers; dividing the historical features corresponding to each merchant reference tier into dimensions according to the merchant rating dimension to obtain historical dimension features, which correspond to target dimension features; calculating the normal distribution curve of the historical dimension features and using the feature values ​​of the historical dimension features that are at preset distribution positions in the normal distribution curve as the dimension reference values ​​of the target dimension features corresponding to the historical dimension features; and summarizing the dimension reference values ​​of the target dimension features corresponding to all historical dimension features to obtain the basic reference value.

[0009] According to an embodiment of this application, pre-training a rating classification model includes: repeatedly training a preset neural network architecture using a pre-acquired training sample set, so that the neural network architecture outputs a training control value corresponding to the intermediate data period of the sample data in the training sample set, and obtaining the training rating data of the sample data in the current data period based on the training control value; wherein, the current data period is the next period after the intermediate data period; the training rating data is used to characterize the operational level that a single merchant described by the sample data in the current data period should be in during the training process of the neural network architecture; the training control value is used to characterize the control data on which the neural network architecture obtains the operational level during the training process; the control data corresponds one-to-one with the merchant rating dimensions involved in obtaining the operational level; training is stopped when the total loss value of the neural network architecture is lower than a preset loss threshold, and the optimal neural network architecture is used as the rating classification model.

[0010] According to an embodiment of this application, obtaining the total loss value includes: calculating a control loss value based on training control values ​​and preset control value labels; calculating a level loss value based on training level data and preset level labels; the control loss value is used to characterize the ability of the control module of the level classification model to output basic control values; the control loss value is used to characterize the ability of the level classification module of the level classification model to output level data; assigning control weights and classification weights to the control loss value and the level loss value respectively; and performing a weighted summation of the control loss value and the level loss value based on the control weights and the classification weights to obtain the total loss value.

[0011] According to an embodiment of this application, the data analysis model is a large-scale model. The large-scale model generates an operational report corresponding to the target merchant, containing audit conclusions and business recommendations, based on grade data and target dimension features. This includes: obtaining the target merchant's horizontal ranking information based on grade data and business information based on target dimension features, under the constraint of preset prompts; cross-comparing the horizontal ranking information and business information to obtain the business indicator gap between the target merchant and similar merchants, and generating business indicator improvement data based on this gap; generating audit text for the horizontal ranking information and business improvement text for the business indicator improvement data using natural language generation technology; combining the audit texts into an audit conclusion and filling the business improvement text as business recommendations into a preset report template to form an operational report.

[0012] According to the embodiments of this application, data extraction is performed on the pre-acquired operational data of the target merchant in the current operating cycle to obtain effective operational data, including: obtaining operational data from the merchant database of the target merchant through a preset application programming interface; cleaning the operational data to remove duplicate data and non-critical data to obtain standard operational data; and filtering out data related to the merchant rating dimension from the standard operational data as effective operational data.

[0013] The second aspect of this application provides a merchant operation level review and analysis device, comprising: a data extraction module for extracting operational data of a target merchant in the current operating cycle from pre-acquired data to obtain effective operational data; a feature extraction module for extracting features from the effective operational data to obtain target dimension features related to merchant rating dimensions, wherein the merchant rating dimensions include at least one of the following: transaction project net profit, transaction settlement net profit, handling fee net profit, total net profit, transaction amount, and number of transactions; a level classification module for classifying the target merchant into merchant levels based on the target dimension features and pre-acquired baseline reference values ​​to obtain level data characterizing the operational status of the target merchant; wherein the baseline reference values ​​are obtained by clustering preset historical operational data, and the historical operational data are the operational data generated by the merchant group, including the target merchant, in the previous operating cycle of the current operating cycle; and a report generation module for generating an operational report corresponding to the target merchant with review conclusions and business recommendations based on the level data and target dimension features using a pre-trained data analysis model.

[0014] A third aspect of this application provides an electronic device comprising: one or more processors; and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors perform the method described above.

[0015] A fourth aspect of this application also provides a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, cause the processor to perform the methods described above.

[0016] The fifth aspect of this application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0017] According to the merchant operation level review and analysis method, device, equipment, medium, and product provided in this application, after extracting effective operational data, feature extraction is performed to obtain target dimension features. Based on these target dimension features and basic reference values, merchant levels are classified to obtain level data. An operational report is generated based on this level data using a data analysis model. Because the operational data of the target merchant within the current operational cycle is extracted first, a small volume of effective operational data is obtained, thus reducing data input for subsequent operations and improving the data targeting of the level review, thereby increasing the accuracy of merchant review and classification. Since the merchant level is classified by comparing it with pre-obtained basic reference values, and these basic reference values ​​are obtained through... This system is derived by clustering historical operational data from a merchant group, including the target merchant. This links the merchant's classification to the historical data of the merchant group to which it belongs, allowing the classified merchant level to intuitively reflect the target merchant's operational status and ranking within the group. This improves the stickiness and relevance of the classification of merchants within the enterprise or group. Furthermore, by using a pre-trained data analysis model to generate operational reports with review conclusions and operational suggestions corresponding to the target merchant based on the classification data and target dimension characteristics, the system can not only automatically output the target merchant's operational level but also provide operational suggestions for the target merchant, further enhancing the model's intelligence in assisting decision-making. Attached Figure Description

[0018] The above-mentioned contents, other objects, features and advantages of this application will become clearer from the following description of embodiments with reference to the accompanying drawings, in which:

[0019] Figure 1 The illustrations schematically depict application scenarios of the merchant operation level review and analysis method, apparatus, equipment, media, and program products according to embodiments of this application.

[0020] Figure 2 A flowchart illustrating the merchant operation level review and analysis method according to an embodiment of this application is shown in the schematic diagram.

[0021] Figure 3 This illustration shows a schematic diagram of the data processing flow involved in the merchant operation level review and analysis method according to an embodiment of this application;

[0022] Figure 4 This illustration schematically shows the functional modules involved in the merchant operation level review and analysis method according to an embodiment of this application, and the related data flow diagram.

[0023] Figure 5 This schematic diagram illustrates the structural block diagram of a merchant operation level review and analysis device according to an embodiment of this application;

[0024] Figure 6 A block diagram schematically illustrates an electronic device suitable for implementing a merchant operation level review and analysis method according to an embodiment of this application. Detailed Implementation

[0025] The embodiments of this application will now be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of this application. In the following detailed description, numerous specific details are set forth to provide a thorough understanding of the embodiments of this application for ease of explanation. However, it will be apparent that one or more embodiments may be implemented without these specific details. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concepts of this application.

[0026] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. The terms “comprising,” “including,” etc., as used herein indicate the presence of the stated features, steps, operations, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, or components.

[0027] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein are to be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0028] When using expressions such as "at least one of A, B and C", they should generally be interpreted in accordance with the meaning that is commonly understood by those skilled in the art (e.g., "a system having at least one of A, B and C" should include, but is not limited to, a system having A alone, a system having B alone, a system having C alone, a system having A and B, a system having A and C, a system having B and C, and / or a system having A, B and C, etc.).

[0029] It should be noted that in the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, it does not mean that the applicant has used or necessarily used the solution.

[0030] In the technical solution of this application, the collection, storage, use, processing, transmission, provision, disclosure and application of user personal information all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0031] In the technical solution of this application, the acquisition, collection, storage, use, processing, transmission, provision, disclosure and application of data all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0032] Figure 1 The illustration shows an application scenario diagram of the merchant operation level review and analysis method according to an embodiment of this application.

[0033] like Figure 1 As shown, application scenario 100 according to this embodiment may include terminal devices 101, 102, and 103, network 104, and server 105. Network 104 is used as a medium to provide a communication link between terminal devices 101, 102, and 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables, etc.

[0034] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social media platform software, etc. (for example only).

[0035] Terminal devices 101, 102, and 103 can be various electronic devices with displays and web browsing capabilities, including but not limited to smartphones, tablets, laptops, and desktop computers.

[0036] Server 105 can be a server that provides various services, such as a backend management server that supports websites browsed by users using terminal devices 101, 102, and 103 (for example only). The backend management server can analyze and process data such as received user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices.

[0037] It should be noted that the merchant operation level review and analysis method provided in this application embodiment can generally be executed by server 105. Correspondingly, the merchant operation level review and analysis device provided in this application embodiment can generally be located in server 105. The merchant operation level review and analysis method provided in this application embodiment can also be executed by a server or server cluster that is different from server 105 and capable of communicating with terminal devices 101, 102, 103 and / or server 105. Correspondingly, the merchant operation level review and analysis device provided in this application embodiment can also be located in a server or server cluster that is different from server 105 and capable of communicating with terminal devices 101, 102, 103 and / or server 105.

[0038] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0039] The following will be based on Figure 1 The described scene, through Figures 2-4 The merchant operation level review and analysis method of the application embodiment is described in detail.

[0040] Figure 2 A flowchart illustrating the merchant operation level review and analysis method according to an embodiment of this application is shown.

[0041] like Figure 2 As shown, the merchant operation level review and analysis method in this embodiment includes operations S210 to S250.

[0042] In operation S210, data extraction is performed on the pre-acquired operational data of the target merchant in the current operating cycle to obtain effective operational data. In operation S220, feature extraction is performed on the effective operational data to obtain target dimension features related to merchant rating dimensions. The merchant rating dimensions include at least one of the following: transaction project net profit, transaction settlement net profit, handling fee net profit, total net profit, transaction amount, and number of transactions. In operation S230, the target merchant is classified into merchant levels according to the target dimension features and the pre-acquired basic reference values ​​to obtain level data that characterizes the operational status of the target merchant. The basic reference values ​​are obtained by clustering preset historical operational data, which is the operational data generated by the merchant group, including the target merchant, in the previous operating cycle of the current operating cycle. In operation S240, an operational report with audit conclusions and business suggestions corresponding to the target merchant is generated based on the level data and target dimension features using a pre-trained data analysis model.

[0043] As an example, a pre-set data acquisition engine is used to obtain operational data of all relevant merchants under a group or enterprise. Merchants awaiting review are selected as target merchants. This data acquisition engine is a pre-set program or functional component that connects to the group or enterprise's data system via an API (Application Programming Interface), facilitating efficient and quick acquisition of the business data of all associated merchants under the group or enterprise. After obtaining the operational data, to reduce the overall computational load of subsequent operations, effective operational data extraction is performed first. This reduces the overall data computation load of subsequent operations and improves the targeting of operational data classification. After obtaining effective operational data, a feature extraction module is used to extract target dimension features. These target dimension features are those associated with the merchant rating dimension, such as at least one of the following: transaction project net profit, transaction settlement net profit, handling fee net profit, total net profit, transaction amount, and transaction number. Then, the merchant rating is determined based on these target dimension features and the baseline comparison value. The grading process can be executed by a pre-set program. However, considering that the program can only perform calculations between data, it still needs to rely on human preset or calculation of basic reference values. Therefore, in this example, a neural network model can be used to cluster the preset historical operating data to obtain basic reference values. This historical operating data is the operating data generated by the merchant group, including the target merchant, in the previous operating cycle of the current operating cycle. This merchant group can be all merchants under the group or enterprise, or it can be some merchants in the group or enterprise that meet certain conditions. There are no restrictions here. Based on all the operating data generated by the merchant group related to the target merchant, an appropriate basic reference value is obtained. Then, based on the basic reference value and the target dimension characteristics, the grade data that is appropriate to the operating level of the target merchant is calculated. Subsequently, the data analysis model generates corresponding audit conclusions and business suggestions based on the grade data, thereby forming the operating report of the target merchant.

[0044] Based on this, the operational data of the target merchant within the current operating cycle is extracted first to obtain a small volume of effective operational data. This reduces the data input for subsequent operations and improves the data targeting of the rating review, thereby increasing the accuracy of merchant rating. The merchant rating is then determined by comparing the data with a pre-obtained baseline reference value, which is obtained by clustering historical operational data of the merchant group, including the target merchant. This links the merchant rating with the historical data of the merchant group, improving the accuracy and adaptability of the merchant rating classification. The resulting merchant rating directly reflects the target merchant's operational status and business performance ranking within the group, enhancing the stickiness and generalization ability of the enterprise in rating its merchants. A pre-trained data analysis model generates an operational report with review conclusions and business recommendations corresponding to the target merchant based on the rating data and target dimension characteristics. This not only automatically outputs the target merchant's operational rating but also provides operational recommendations, further improving the intelligence level of decision support.

[0045] In this embodiment, the classification module in the pre-trained classification model classifies target merchants into different levels based on target dimension features and pre-acquired baseline reference values, thereby obtaining level data to characterize the operational status of the target merchants. This includes: obtaining dimension reference values ​​corresponding to the target dimension features from the baseline reference values, whereby the dimension reference values ​​represent the reference data contained in the baseline reference values ​​that correspond one-to-one with the merchant rating dimensions; performing weighted calculations based on the dimension values ​​in the target dimension features, the dimension reference values, and the dimension weights corresponding one-to-one with the target dimension features to obtain dimension weights, whereby the dimension weights are either the first dimension weight or the second dimension weight; summing the dimension weights corresponding to each target dimension feature to obtain merchant indicator data; and determining the level data corresponding to the merchant indicator data in a preset indicator level mapping relationship table.

[0046] As an example, this grading model is a neural network model, where the grading module is part of the neural network model of the training number. It can be a branch model constituting the grading model or a functional layer of the grading model; there are no restrictions, as long as it can satisfy the process of grading target merchants based on target dimension features. Specifically, when this grading module grades target merchants, the basic reference values ​​include numerous reference data. Each target dimension feature corresponding to a merchant rating dimension has a set of reference data, which are dimension reference values. For example, the dimension reference values ​​corresponding to the six dimensions—transaction project net profit, transaction settlement net profit, handling fee net profit, total net profit, transaction amount, and transaction number—can be denoted as s1, s2, s3, s4, s5, and s6, respectively. The dimension weights are obtained by weighting the dimension values ​​in the target dimension features, the dimension reference values, and the dimension weights corresponding to the target dimension features. The dimension weight is either the first dimension weight or the second dimension weight. For example, the dimension values ​​of the six target dimension features corresponding to the six merchant rating dimensions are NP1, NP2, NP3, NP4, NP5, and NP6, and the dimension weights corresponding to these six target dimension features are r1, r2, r3, r4, r5, and r6, respectively. A weighted calculation is then performed to obtain the dimension weights v1, v2, v3, v4, v5, and v6. After obtaining the dimension weights, these six dimension weights are added together to obtain the merchant indicator data v, where v = v1 + v2 + v3 + v4 + v5 + v6. Then, the corresponding level data for the merchant indicator data is determined in a preset indicator level mapping table. This indicator level mapping table is a fixed, optimal indicator level mapping table generated during the initial training of the level classification model. For example, if the merchant indicator data v is in the range of 1-10, then the level data is level one; if the merchant indicator data v is in the range of 10-20, then the level data is level two, and so on.

[0047] Merchants are classified into different levels using a pre-trained classification model. Based on this process, dimensional weights are obtained from the baseline reference data. These weights are then summed to obtain merchant indicator data. Finally, the merchant level data corresponding to this indicator data is determined based on an indicator level mapping table. This allows for a rigorous calculation of merchant indicator data based on target dimensional features, reference data, and dimensional weights. The customer indicator data is objectively calculated based on baseline reference values ​​obtained from historical operational data. This customer indicator data objectively and rigorously reflects the target merchant's operational ranking / level within the enterprise or group's customer base. This enables the enterprise or group to fully understand the merchant's operational status and performance within the group, improving classification accuracy and enhancing the intelligence of decision-making support.

[0048] In this embodiment, when the dimension value is less than or equal to the dimension reference value, the dimension weight is the first dimension weight; wherein, calculating the first dimension weight includes: dividing the dimension value by the dimension reference value to obtain the operational reference ratio value; multiplying the operational reference ratio value by the dimension weight corresponding one-to-one with the target dimension feature to obtain the first dimension weight; when the dimension weight is greater than the dimension reference value, the dimension weight is the second dimension weight, and the second dimension weight is the first dimension weight plus one.

[0049] The dimensional comparison values ​​can be denoted as s1, s2, s3, s4, s5, and s6. The dimensional values ​​of the six target dimensional features corresponding to these six merchant rating dimensions are NP1, NP2, NP3, NP4, NP5, and NP6. The dimensional weights corresponding to these six target dimensional features are r1, r2, r3, r4, r5, and r6, respectively. Weighted calculations are then performed to obtain the dimensional weights v1, v2, v3, v4, v5, and v6. Taking v1 as an example... This statement determines whether the dimension value NP1 corresponding to the net profit of the transaction project is greater than the dimension comparison value s1 corresponding to the net profit of the transaction project. If it is greater, v1 is calculated according to the formula after the colon ":", that is... Its dimension weight v1 is obtained by dividing the dimension value NP1 by the dimension comparison value s1, resulting in the operational comparison ratio value; then, this operational comparison ratio value is multiplied by the dimension weight r1 that corresponds one-to-one with the target dimension feature. This is the weight of the first dimension; if it is not greater than, i.e., less than or equal to, then v1 is calculated according to the formula before the colon ":", i.e. ,at this time, Obviously, this is the first dimension weight plus one; the calculation methods for v2, v3, v4, v5, and v6 are the same as v1, and will not be repeated here. After calculating the dimension weights v1, v2, v3, v4, v5, and v6, the weights of these six dimensions are added together to obtain the merchant indicator data v, where v = v1 + v2 + v3 + v4 + v5 + v6. It should be noted that the above... It indicates multiplication / multiplying by / multiplying.

[0050] Based on this, when the dimension value is greater than the dimension reference value, its dimension weight is increased by 1 relative to the case where the dimension value is not greater than the dimension reference value. This allows for the division of dimension weights into different gradients. Then, based on the relationship between the dimension value and the dimension reference value, different gradients of merchant indicator data are divided, thereby distinguishing the merchant level data. This objective and rigorous division of merchant levels based on dimension weights improves the accuracy and rigor of merchant ranking and classification.

[0051] In this embodiment, the baseline reference value is obtained by clustering preset historical operational data through the reference module in the rating model. This includes: retrieving historical operational data related to the target merchant and extracting features from the historical operational data to obtain historical features; clustering the historical features based on a preset expected ratio of levels to obtain merchant reference levels; dividing the historical features corresponding to each merchant reference level into dimensions according to the merchant rating dimension to obtain historical dimension features, which correspond to the target dimension features; calculating the normal distribution curve of the historical dimension features, and using the feature values ​​of the historical dimension features that are at preset distribution positions in the normal distribution curve as the dimension reference values ​​of the target dimension features corresponding to the historical dimension features; and summarizing the dimension reference values ​​of the target dimension features corresponding to all historical dimension features to obtain the baseline reference value.

[0052] As an example, the process of obtaining the basic reference value is executed by the reference module in the aforementioned grading model. This reference module can be a branch model of the grading model or a functional layer of the grading model; there are no restrictions here. In this example, when obtaining the basic reference value, all historical operational data related to the target merchant are first retrieved, and features are extracted from the historical operational data to obtain historical features. Then, based on the preset expected proportion of the grading level, the historical features are clustered to obtain the merchant reference level. This merchant reference level is the level data that the target merchant may be rated as. For example, if the merchant reference level is divided into five levels, the expected proportions of the five levels can be divided into Level 1 1%, Level 2 5%, Level 3 34%, Level 4 35%, and Level 5 25%, respectively. This expected proportion is only for example and represents the proportion of merchants of this level among all merchants participating in the merchant grading.

[0053] Clustering is performed according to this ratio to obtain merchant comparison levels. These merchant comparison levels include the historical characteristics of each merchant comparison registration under the cluster division. There are no restrictions on the clustering method, as long as it can be divided into five categories proportionally. Then, the historical characteristics under each merchant comparison level are divided into dimensions to obtain historical dimension features. The historical dimension features need to correspond to the target dimension features mentioned above. Then, the normal distribution curve of the historical dimension features is calculated, and the feature values ​​of the historical dimension features that are located at the preset distribution position in the normal distribution curve are used as the dimension comparison values ​​of the target dimension features corresponding to the historical dimension features. For example, the feature values ​​of the historical dimension features distributed in the +5% of the normal distribution curve can be used as the dimension comparison values ​​corresponding to their target dimension features. Then, the dimension comparison values ​​of the target dimension features corresponding to all historical dimension features are summarized to obtain the basic comparison values.

[0054] Therefore, by using techniques such as clustering and normal distribution curve division, basic reference values ​​are obtained, thus accurately deriving basic reference values ​​that correspond to the expected target merchants at each level, thereby improving the accuracy and rigor of the level classification.

[0055] In this embodiment, pre-training the rating classification model includes: repeatedly training a preset neural network architecture using a pre-acquired training sample set, so that the neural network architecture outputs a training control value corresponding to the intermediate data period of the sample data in the training sample set, and obtaining the training rating data of the sample data in the current data period based on the training control value; wherein, the current data period is the next period after the intermediate data period; the training rating data is used to characterize the operational level that a single merchant described by the sample data in the current data period should be in during the training process of the neural network architecture; the training control value is used to characterize the control data on which the neural network architecture obtains the operational level during the training process; the control data corresponds one-to-one with the merchant rating dimensions involved in obtaining the operational level; training stops when the total loss value of the neural network architecture is lower than a preset loss threshold, and the optimal neural network architecture is used as the rating classification model.

[0056] As an example, this ranking model is essentially a pre-trained neural network model. This neural network model is pre-trained based on a preset neural network architecture. The training process is very similar to the operation flow of the application stage of the aforementioned ranking model, except that its input is data from the training sample set, and it includes an additional step of calculating a loss function based on the output. The neural network model is then optimized based on the feedback of this loss function. For example, the data in the intermediate data period of the sample data in the training sample set is equivalent to the historical operation data in the application stage, and the training control value corresponding to the data in the intermediate data period of the sample data is equivalent to the basic control value obtained in the application stage. The training ranking data in the current data period is equivalent to the effective operation data in the application stage. The training ranking data generated based on the training ranking data and the training control value is equivalent to the ranking data in the application stage. The training continues, and the corresponding total loss value is calculated until the total loss value of the neural network architecture is lower than the preset loss threshold. At this point, training stops, and the optimal neural network architecture is used as the ranking model.

[0057] This process enables the trained grading model to automatically generate a reasonable baseline reference value based on the operational data generated in the previous operational cycle. Based on this baseline reference value and the operational data of the current operational cycle, it can then classify the merchant grading data into a rigorous grading system, thereby improving the accuracy and rigor of merchant grading and enhancing the intelligence level of auxiliary decision-making in merchant grading.

[0058] In this embodiment, obtaining the total loss value includes: calculating a control loss value based on the training control value and a preset control value label; calculating a level loss value based on the training level data and a preset level label; the control loss value is used to characterize the ability of the control module of the level classification model to output basic control values; the control loss value is used to characterize the ability of the level classification module of the level classification model to output level data; assigning control weights and classification weights to the control loss value and the level loss value respectively; and performing a weighted summation of the control loss value and the level loss value based on the control weights and the classification weights to obtain the total loss value.

[0059] The total loss value consists of two parts: one part is the level loss value generated during the training phase of the comparison module in the level classification model, and the other part is the comparison loss value corresponding to the level classification module in the level classification model. The two are weighted and summed to obtain the total loss value. Based on this, the trained level classification model not only focuses on the accuracy of level classification, but also on the accuracy of generating basic comparison values ​​in the process. The two work together to further improve the accuracy and rigor of merchant level classification.

[0060] In this embodiment, the data analysis model is a large-scale model. This model generates an operational report corresponding to the target merchant, containing review conclusions and business recommendations, based on grade data and target dimension features. This includes: obtaining the target merchant's horizontal ranking information based on grade data and business information based on target dimension features, under the constraint of preset prompts; cross-comparing the horizontal ranking information and business information to determine the gap in business indicators between the target merchant and similar merchants, and generating business indicator improvement data based on this gap; using natural language generation technology to generate review text for the horizontal ranking information and business improvement text for the business indicator improvement data; combining the review text into a review conclusion and filling the business improvement text as business recommendations into a preset report template to form the operational report.

[0061] As an example, the data analysis model used to generate the operation report is a large model, specifically an artificial intelligence large model. This refers to a type of artificial intelligence model with a large number of parameters, constructed from artificial neural networks. Such large AI models typically include large language models, visual models, multimodal models, and basic science models. In this embodiment, the large model includes machine text models with different functions; this machine text model is the text large model. When generating an operation report with review conclusions and business suggestions corresponding to the target merchant based on ranking data and target dimension features using this text large model, firstly, under the constraint of preset prompts, the target merchant's ranking information is obtained based on the ranking data, and then business information, such as the merchant's business items, is obtained based on the target dimension features. What is it, and what is the profit? Then, based on the horizontal ranking information and business information, a cross-comparison is performed to obtain the business indicator gap between the target merchant and similar merchants. This business indicator gap is how much merchandise needs to be sold to reach the industry-leading level. Based on this business indicator gap, business indicator improvement data is generated. Then, natural language generation technology is used to generate review text based on the horizontal ranking information, such as the merchant's ranking within the enterprise or group, and the merchant's operational level ranking in its field. Then, business improvement text is generated based on the business indicator improvement data, such as operational suggestions on how to improve operational level. Finally, the review text is grouped into the review conclusion, and the business improvement text is filled into the preset report template as business suggestions to form an operational report.

[0062] Based on this, it can not only accurately output information about merchant level classification, but also provide reasonable and targeted suggestions for the merchant, improve the relevance and breadth of the information output by the model for merchant review and evaluation, and thus improve the intelligence level of the model in assisting enterprise decision-making.

[0063] In this embodiment, the operation data of the target merchant in the current operation cycle is extracted in advance to obtain effective operation data, including: obtaining the operation data in the merchant database of the target merchant through a preset application interface; cleaning the operation data to remove duplicate data and non-critical data to obtain standard operation data; and filtering out data related to the merchant rating dimension from the standard operation data as effective operation data.

[0064] As an example, when obtaining valid operational data, a pre-set data acquisition engine is used to retrieve the operational data of all relevant merchants under the group or enterprise from the merchant database. The merchant to be reviewed is selected as the target merchant. This data acquisition engine is a pre-set program or functional component connected to the merchant database via API, facilitating efficient and quick retrieval of the business data of all associated merchants under the group or enterprise. After obtaining the operational data, it is necessary to perform data cleaning to remove duplicate and non-critical data, obtaining standard operational data. This reduces the overall computational load of subsequent operational steps. Data related to the merchant rating dimensions is then selected from the standard operational data as valid operational data, further reducing the overall computational load of subsequent operational steps and improving the relevance of operational data classification. This valid operational data involves merchant rating dimensions, which include at least one of the following: transaction project net profit, transaction settlement net profit, handling fee net profit, total net profit, transaction amount, and number of transactions. In a more specific example, the transaction project net profit can be... The target dimension feature related to this dimension can be denoted as NP1: Merchant monthly loan net profit refers to the acquiring bank's merchant loan profit minus costs; Transaction settlement net profit refers to the merchant's monthly settlement account deposit net profit, specifically the acquiring bank's merchant settlement account deposit income minus costs; Transaction settlement net profit refers to the monthly transaction fee net profit, specifically the acquiring bank's receivable transaction fees minus the acquiring bank's payable channel costs; Total net profit refers to the merchant's monthly direct contribution net profit, which is the above-mentioned transaction fee net profit + merchant settlement account deposit net profit + merchant loan net profit; Transaction amount refers to the monthly transaction amount or annual transaction amount, specifically the merchant's transaction amount within a time period; Transaction number refers to the monthly transaction number or annual transaction number, specifically the number of transactions made by the merchant within a certain time period; Transaction number refers to the monthly transaction number or annual transaction number; Transaction number refers to the merchant's transaction number within a certain time period; Transaction number refers to the target dimension feature related to this dimension.

[0065] In this way, routine operational data is cleaned and filtered to obtain the most effective operational data for merchant rating dimensions, thereby reducing the amount of data input to the subsequent model. While reducing the model's data input and computational load, the model's generalization ability and the effectiveness of data output are improved.

[0066] As described above, the merchant operation level review and analysis method provided in this application reduces the data input for subsequent operations and improves the data targeting of the level review, thereby improving the accuracy of merchant review and classification. Furthermore, it allows the merchant classification to be correlated with the historical data of the merchant group to which it belongs, enabling the classified merchant level to intuitively reflect the target merchant's operational status and business performance ranking within the group. This enhances the stickiness and relevance of the classification of merchants within an enterprise or group. Moreover, because the pre-trained data analysis model generates an operational report with review conclusions and operational suggestions corresponding to the target merchant based on the level data and target dimension characteristics, it can not only automatically output the target merchant's operation level but also provide operational suggestions for the target merchant, further improving the model's intelligence level in assisting decision-making.

[0067] Based on the aforementioned merchant operation level review and analysis method, this application also provides a merchant operation level review and analysis device. The following will combine... Figure 5 The device is described in detail.

[0068] Figure 5 The diagram illustrates the structure of a merchant operation level review and analysis device according to an embodiment of this application.

[0069] like Figure 5 As shown, the merchant operation level review and analysis device 500 in this embodiment includes a data extraction module 510, a feature extraction module 520, a level classification module 530, and a report generation module 540.

[0070] The data extraction module 510 includes: a data acquisition unit, used to acquire operational data from the merchant database of the target merchant through a preset application programming interface; a data cleaning unit, used to clean the operational data to remove duplicate and non-critical data to obtain standard operational data; and a data filtering unit, used to filter out data related to the merchant rating dimension from the standard operational data as valid operational data.

[0071] The rating module 530 uses a rating classification module in a pre-trained rating model to classify target merchants according to target dimension features and pre-acquired baseline comparison values, obtaining rating data to characterize the operational status of target merchants. This includes: obtaining dimension comparison values ​​corresponding to the target dimension features from the baseline comparison values, where each dimension comparison value represents the comparison data contained in the baseline comparison values ​​that corresponds one-to-one with the merchant rating dimensions; performing weighted calculations based on the dimension values, dimension comparison values, and dimension weights corresponding to the target dimension features to obtain dimension weights, which are either first dimension weights or second dimension weights; summing the dimension weights corresponding to each target dimension feature to obtain merchant indicator data; and determining the rating data corresponding to the merchant indicator data in a pre-set indicator rating mapping table. Wherein, when the dimension value is less than or equal to the dimension comparison value, the dimension weight is the first dimension weight. Calculating the first dimension weight includes: dividing the dimension value by the dimension comparison value to obtain an operational comparison ratio value; and... The first dimension weight is obtained by multiplying the operational comparison ratio value by the dimension weight corresponding to the target dimension feature. If the dimension weight is greater than the dimension comparison value, the dimension weight becomes the second dimension weight, which is the first dimension weight plus one. The basic comparison value is obtained by clustering the preset historical operational data through the comparison module in the grading model. This includes: retrieving historical operational data related to the target merchant and extracting features from the historical operational data to obtain historical features; clustering the historical features based on the preset expected proportion of the tier to obtain the merchant comparison tier; dividing the historical features corresponding to each merchant comparison tier according to the merchant rating dimension to obtain historical dimension features, which correspond to the target dimension features; calculating the normal distribution curve of the historical dimension features, and using the feature values ​​of the historical dimension features that are in the preset distribution position in the normal distribution curve as the dimension comparison values ​​of the target dimension features corresponding to the historical dimension features; and summing up the dimension comparison values ​​of the target dimension features corresponding to all historical dimension features to obtain the basic comparison value.

[0072] The grading model used in the grading module 530 is pre-trained. Pre-training the grading model includes: repeatedly training a pre-acquired training sample set with a preset neural network architecture, so that the neural network architecture outputs a training control value corresponding to the intermediate data period of the sample data in the training sample set, and obtaining the training grade data of the sample data in the current data period based on the training control value; wherein the current data period is the period following the intermediate data period; the training grade data is used to characterize the operational level that a single merchant should be in the current data period as described by the sample data during the training process of the neural network architecture; the training control value is used to characterize the control data on which the neural network architecture obtains the operational level during the training process; the control number... The model is designed to correspond one-to-one with the merchant rating dimensions involved in obtaining the operational level; training stops when the total loss value of the neural network architecture is lower than a preset loss threshold, and the optimal neural network architecture is used as the level classification model; obtaining the total loss value includes: calculating the control loss value based on the training control value and the preset control value label, and calculating the level loss value based on the training level data and the preset level label. The control loss value is used to characterize the ability of the control module of the level classification model to output the basic control value, and the control loss value is used to characterize the ability of the level classification module of the level classification model to output the level data; control weights and classification weights are assigned to the control loss value and the level loss value respectively, and the control loss value and the level loss value are weighted and summed based on the control weights and classification weights to obtain the total loss value.

[0073] The data analysis model used in the report generation module 540 is a large model. This large model generates an operational report corresponding to the target merchant, containing audit conclusions and business recommendations, based on the level data and the target dimension features. This includes: obtaining the target merchant's horizontal ranking information based on the level data and business information based on the target dimension features, under the constraint of preset prompts; cross-comparing the horizontal ranking information and business information to determine the gap in business indicators between the target merchant and similar merchants, and generating business indicator improvement data based on this gap; generating audit text for the horizontal ranking information and business improvement text for the business indicator improvement data using natural language generation technology; combining the audit text into an audit conclusion and filling the business improvement text as business recommendations into a preset report template to form the operational report.

[0074] According to embodiments of this application, any multiple modules among the data extraction module 510, feature extraction module 520, classification module 530, and report generation module 540 can be merged into one module, or any one of these modules can be split into multiple modules. Alternatively, at least some of the functions of one or more of these modules can be combined with at least some of the functions of other modules and implemented in one module. According to embodiments of this application, at least one of the data extraction module 510, feature extraction module 520, classification module 530, and report generation module 540 can be at least partially implemented as hardware circuitry, such as a field-programmable gate array (FPGA), a programmable logic array (PLA), a system-on-a-chip, a system-on-a-substrate, a system-on-package, an application-specific integrated circuit (ASIC), or implemented in hardware or firmware by any other reasonable means of integrating or packaging the circuitry, or implemented in software, hardware, or firmware, or in any suitable combination of any of these three implementation methods. Alternatively, at least one of the data extraction module 510, feature extraction module 520, grading module 530, and report generation module 540 can be at least partially implemented as a computer program module, which can perform corresponding functions when the computer program module is run.

[0075] Figure 6 A block diagram schematically illustrates an electronic device suitable for implementing a merchant operation level review and analysis method according to an embodiment of this application.

[0076] like Figure 6 As shown, an electronic device 600 according to an embodiment of this application includes a processor 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage portion 608 into a random access memory (RAM) 603. The processor 601 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or an associated chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 601 may also include onboard memory for caching purposes. The processor 601 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of this application.

[0077] RAM 603 stores various programs and data required for the operation of electronic device 600. Processor 601, ROM 602, and RAM 603 are interconnected via bus 604. Processor 601 executes various operations of the method flow according to embodiments of this application by executing programs in ROM 602 and / or RAM 603. It should be noted that the programs may also be stored in one or more memories other than ROM 602 and RAM 603. Processor 601 may also execute various operations of the method flow according to embodiments of this application by executing programs stored in said one or more memories.

[0078] According to embodiments of this application, the electronic device 600 may further include an input / output (I / O) interface 605, which is also connected to a bus 604. The electronic device 600 may also include one or more of the following components connected to the I / O interface 605: an input section 606 including a keyboard, mouse, etc.; an output section 607 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 910 is also connected to the I / O interface 605 as needed. A removable medium 611, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 610 as needed so that computer programs read from it can be installed into the storage section 608 as needed.

[0079] This application also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or it may exist independently and not assembled into the device / apparatus / system. The computer-readable storage medium carries one or more programs, which, when executed, implement the method according to the embodiments of this application.

[0080] According to embodiments of this application, the computer-readable storage medium can be a non-volatile computer-readable storage medium, such as including but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, according to embodiments of this application, the computer-readable storage medium may include ROM 602 and / or RAM 603 and / or one or more memories other than ROM 602 and RAM 603 described above.

[0081] Embodiments of this application also include a computer program product comprising a computer program containing program code for performing the methods shown in the flowchart. When the computer program product is run on a computer system, the program code is used to enable the computer system to implement the item recommendation method provided in the embodiments of this application.

[0082] When the computer program is executed by the processor 601, it performs the functions defined in the system / apparatus of this application embodiment. According to the embodiments of this application, the systems, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0083] In one embodiment, the computer program may rely on a tangible storage medium such as an optical storage device or a magnetic storage device. In another embodiment, the computer program may also be transmitted and distributed in the form of signals over a network medium, and downloaded and installed via the communication section 609, and / or installed from the removable medium 611. The program code contained in the computer program can be transmitted using any suitable network medium, including but not limited to: wireless, wired, etc., or any suitable combination thereof.

[0084] In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 609, and / or installed from the removable medium 611. When the computer program is executed by the processor 601, it performs the functions defined in the system of this application embodiment. According to the embodiments of this application, the systems, devices, apparatuses, modules, units, etc., described above can be implemented by computer program modules.

[0085] According to embodiments of this application, program code for executing the computer programs provided in the embodiments of this application can be written in any combination of one or more programming languages. Specifically, these computational programs can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, languages ​​such as Java, C++, Python, "C", or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0086] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0087] Those skilled in the art will understand that the features described in the various embodiments and / or claims of this application can be combined or combined in various ways, even if such combinations or combinations are not explicitly described in this application. In particular, the features described in the various embodiments and / or claims of this application can be combined or combined in various ways without departing from the spirit and teachings of this application. All such combinations and / or combinations fall within the scope of this application.

[0088] The embodiments of this application have been described above. However, these embodiments are merely illustrative and not intended to limit the scope of this application. Although various embodiments have been described above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. The scope of this application is defined by the appended claims and their equivalents. Various substitutions and modifications can be made by those skilled in the art without departing from the scope of this application, and all such substitutions and modifications should fall within the scope of this application.

Claims

1. A method for merchant operation level review and analysis, characterized in that, include: Extract effective operational data from the pre-acquired operational data of the target merchants in the current operational cycle; Feature extraction is performed on the effective operational data to obtain target dimension features for merchant rating dimensions. The merchant rating dimensions include at least one of the following: net profit of transaction items, net profit of transaction settlement, net profit of handling fees, total net profit, transaction amount, and number of transactions. Based on the target dimension features and the pre-obtained basic reference values, the target merchants are classified into merchant levels to obtain level data that characterizes the operational status of the target merchants; wherein, the basic reference values ​​are obtained by clustering preset historical operational data, and the historical operational data are the operational data generated by the merchant group including the target merchants in the previous operational cycle of the current operational cycle. An operational report with audit conclusions and business recommendations is generated based on the grade data and the target dimension features using a pre-trained data analysis model.

2. The merchant operation level review and analysis method according to claim 1, characterized in that, The target merchants are classified into different levels by the level classification module in the pre-trained level classification model according to the target dimension features and the pre-acquired basic reference values, so as to obtain level data to characterize the operation status of the target merchants. These include: Obtain the dimension comparison value corresponding to the target dimension feature from the basic comparison value. The dimension comparison value represents the comparison data contained in the basic comparison value that corresponds one-to-one with the merchant rating dimension. The dimension weight is calculated by weighting the dimension value in the target dimension feature, the dimension reference value, and the dimension weight corresponding to the target dimension feature one by one. The dimension weight is either the first dimension weight or the second dimension weight. The merchant metric data is obtained by summing the dimensional weights corresponding to the features of each target dimension; Determine the level data corresponding to the merchant's indicator data in the preset indicator level mapping table.

3. The merchant operation level review and analysis method according to claim 2, characterized in that, When the dimension value is less than or equal to the dimension reference value, the dimension weight is the first dimension weight; wherein, calculating the first dimension weight includes: dividing the dimension value by the dimension reference value to obtain the operation reference ratio value; multiplying the operation reference ratio value by the dimension weight corresponding one-to-one with the target dimension feature to obtain the first dimension weight; When the dimension weight is greater than the dimension reference value, the dimension weight is the second dimension weight, and the second dimension weight is the first dimension weight plus one.

4. The merchant operation level review and analysis method according to claim 3, characterized in that, The baseline reference values ​​are obtained by clustering the preset historical operational data using the reference module in the grading model, including: Retrieve historical operational data related to the target merchant and extract features from the historical operational data to obtain historical features; The historical features are clustered based on a preset hierarchical expected ratio to obtain the merchant comparison level. According to the merchant rating dimensions, the historical characteristics corresponding to each merchant's rating level are divided into dimensions to obtain historical dimension characteristics, which correspond to the target dimension characteristics. Calculate the normal distribution curve of the historical dimension features, and use the feature values ​​of the historical dimension features that are located at a preset distribution position in the normal distribution curve as the dimension comparison values ​​of the target dimension features corresponding to the historical dimension features; The baseline comparison values ​​are obtained by summarizing the dimensional comparison values ​​of the target dimensional features corresponding to all historical dimensional features.

5. The merchant operation level review and analysis method according to claim 4, characterized in that, Pre-training the ranking model includes: A pre-acquired training sample set is used to repeatedly train a preset neural network architecture, so that the neural network architecture outputs a training control value corresponding to an intermediate data period of the sample data in the training sample set, and the training level data of the sample data in the current data period is obtained based on the training control value; wherein, the current data period is the next period after the intermediate data period; the training level data is used to characterize the operational level that a single merchant described by the sample data in the current data period should be in the operational level during the training process of the neural network architecture; the training control value is used to characterize the control data on which the neural network architecture obtains the operational level during the training process; the control data corresponds one-to-one with the merchant rating dimensions involved in obtaining the operational level. Training stops when the total loss value of the neural network architecture falls below a preset loss threshold, and the optimal neural network architecture is used as the ranking model.

6. The merchant operation level review and analysis method according to claim 5, characterized in that, Obtaining the total loss value includes: The control loss value is calculated based on the training control value and the preset control value label, and the level loss value is calculated based on the training level data and the preset level label. The control loss value is used to characterize the ability of the control module of the level division model to output the basic control value, and the control loss value is used to characterize the ability of the level classification module of the level division model to output the level data. Assign control weights and division weights to the control loss value and the grade loss value respectively. Then, perform a weighted summation of the control loss value and the grade loss value based on the control weights and the division weights to obtain the total loss value.

7. The merchant operation level review and analysis method according to claim 6, characterized in that, The data analysis model is a large-scale model, which generates an operational report corresponding to the target merchant, containing audit conclusions and business recommendations, based on the graded data and the target dimension characteristics. This report includes: Under the constraint of preset prompt words, the horizontal ranking information of the target merchant is obtained based on the level data, and business information is obtained based on the target dimension features; By cross-comparing the horizontal ranking information and the business information, the gap between the target merchant and similar merchants in terms of business indicators is obtained, and business indicator improvement data is generated based on the gap in business indicators. Natural language generation technology is used to generate review text for the ranking information and business improvement text for the business indicator improvement data. The audit text group is the audit conclusion, and the business improvement text is the business suggestion. These are then filled into the preset report template to form an operation report.

8. The merchant operation level review and analysis method according to claim 1, characterized in that, Data extraction is performed on the pre-acquired operational data of the target merchants in the current operational cycle to obtain effective operational data, including: The operational data of the target merchant's merchant database is obtained through a preset application programming interface; The operational data is cleaned to remove duplicate and non-critical data, resulting in standard operational data. Data related to the merchant rating dimension is selected from the standard operational data and used as valid operational data.

9. A merchant operation level review and analysis device, comprising: The data extraction module is used to extract operational data from the target merchants in the current operating cycle, which is obtained in advance, to obtain effective operational data; The feature extraction module is used to extract features from the effective operational data to obtain target dimension features related to merchant rating dimensions. The merchant rating dimensions include at least one of the following: transaction project net profit, transaction settlement net profit, handling fee net profit, total net profit, transaction amount, and number of transactions. The rating module is used to classify the target merchants into different ratings based on the target dimension features and the pre-acquired basic reference values, so as to obtain rating data that characterizes the operational status of the target merchants; wherein, the basic reference values ​​are obtained by clustering preset historical operational data, and the historical operational data are the operational data generated by the merchant group including the target merchants in the previous operational cycle in the current operational cycle. The report generation module is used to generate an operational report with audit conclusions and business recommendations corresponding to the target merchant based on the grade data and the target dimension features using a pre-trained data analysis model.

10. An electronic device, comprising: One or more processors; Storage device for storing one or more programs. Wherein, when the one or more programs are executed by the one or more processors, the one or more processors perform the method according to any one of claims 1 to 8.

11. A computer-readable storage medium having stored thereon executable instructions that, when executed by a processor, cause the processor to perform the method according to any one of claims 1 to 8.

12. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1 to 8.

Images