Threshold determination method and device, equipment, storage medium and program product

Abstract

The invention provides a threshold determination method and device, equipment, a storage medium and a program product, and relates to the field of artificial intelligence. The method comprises the following steps: inputting a plurality of business data samples for a target business rule into an unsupervised rule model, and performing feature extraction on the business data samples by the unsupervised rule model to obtain first feature vectors corresponding to a plurality of features and a second feature vector corresponding to a single feature; screening the service data samples based on the first feature vector and the second feature vector to obtain a first abnormal service data sample corresponding to the first feature vector and a second abnormal service data sample corresponding to the second feature vector; and according to the first abnormal business data sample and the second abnormal business data sample, determining a target threshold value corresponding to a target parameter contained in the target business rule, the target threshold value being used as a threshold value of the target parameter in the business rule model. The target threshold value corresponding to the target parameter can be determined more accurately, and resource occupation of the system is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence, and in particular to a threshold determination method and device, equipment, a storage medium and a program product. BACKGROUND

[0002] When making decisions in the fields of marketing, operation and risk control, it is necessary to rely on a business rule model to evaluate the effect of the decisions according to the results of the business rule model, so as to timely adjust and optimize the management strategy.

[0003] At present, the threshold of the business rule model is usually determined based on expert experience. However, in some scenarios, the threshold of the business rule model determined in the above manner is not accurate enough, resulting in resource waste in the system (such as a marketing system) in which the business rule model is located, and affecting work efficiency. SUMMARY

[0004] The present application provides a threshold determination method, device, equipment, storage medium and program product to solve the technical problem that the threshold of the business rule model determined by the current method is not accurate enough, resulting in resource waste in the system in which the business rule model is located, and affecting work efficiency.

[0005] In a first aspect, the present application provides a threshold determination method, comprising:

[0006] obtaining a plurality of business data samples for a target business rule, the target business rule comprising a business rule application target, a definition of the business rule, a target parameter to be determined threshold, and customer attribute information;

[0007] inputting the plurality of business data samples into an unsupervised rule model, the unsupervised rule model performing feature extraction on the business data samples to obtain a first feature vector corresponding to a plurality of features and a second feature vector corresponding to a single feature, the plurality of features being determined according to the business data samples, and the single feature being determined according to the target parameter; filtering the business data samples based on the first feature vector and the second feature vector to obtain first abnormal business data samples corresponding to the first feature vector and second abnormal business data samples corresponding to the second feature vector; and determining a target threshold corresponding to the target parameter according to the first abnormal business data samples and the second abnormal business data samples, the target threshold being used as a threshold of the target parameter in the business rule model.

[0008] In a second aspect, the present application provides a threshold determination device, comprising:

[0009] an obtaining module configured to obtain a plurality of business data samples for a target business rule, the target business rule comprising a business rule application target, a definition of the business rule, a target parameter to be determined threshold, and customer attribute information;

[0010] The processing module is configured to input a plurality of service data samples into an unsupervised rule model, the unsupervised rule model extracts features from the service data samples to obtain a first feature vector corresponding to a plurality of features and a second feature vector corresponding to a single feature, the plurality of features are determined according to the service data samples, and the single feature is determined according to a target parameter; the service data samples are filtered based on the first feature vector and the second feature vector to obtain first abnormal service data samples corresponding to the first feature vector and second abnormal service data samples corresponding to the second feature vector; and the target threshold corresponding to the target parameter is determined according to the first abnormal service data samples and the second abnormal service data samples, and the target threshold is used as a threshold of the target parameter in the service rule model.

[0011] In a third aspect, the present application provides an electronic device, comprising: a processor, and a memory connected with the processor in communication;

[0012] The memory stores computer-executable instructions.

[0013] The processor executes the computer-executable instructions stored in the memory to implement the first aspect and / or various possible implementation manners of the first aspect.

[0014] In a fourth aspect, the present application provides a computer-readable storage medium, the computer-readable storage medium stores computer program instructions, and the computer program instructions are executed to implement the first aspect and / or various possible implementation manners of the first aspect.

[0015] In a fifth aspect, the present application provides a computer program product, comprising a computer program, and the computer program is executed to implement the first aspect and / or various possible implementation manners of the first aspect.

[0016] The threshold determination method, device, equipment, storage medium and program product provided by the present application, by obtaining a plurality of service data samples for a target service rule, the target service rule including a service rule application target, a definition of the service rule, a target parameter to be determined threshold and customer attribute information; inputting the plurality of service data samples into an unsupervised rule model, the unsupervised rule model performing feature extraction on the service data samples to obtain a first feature vector corresponding to a plurality of features and a second feature vector corresponding to a single feature, the plurality of features being determined according to the service data samples, and the single feature being determined according to the target parameter; filtering the service data samples based on the first feature vector and the second feature vector to obtain a first abnormal service data sample corresponding to the first feature vector and a second abnormal service data sample corresponding to the second feature vector; determining a target threshold corresponding to the target parameter according to the first abnormal service data sample and the second abnormal service data sample, the target threshold being used as a threshold of the target parameter in the service rule model. In the present application, the first abnormal service data sample corresponding to the first feature vector and the second abnormal service data sample corresponding to the second feature vector are obtained by filtering through the unsupervised rule model, and then the target threshold corresponding to the target parameter is determined according to the first abnormal service data sample and the second abnormal service data sample, realizing the use of the outlier filtering method in the unsupervised rule model based on the actual service data, and combining the outlier filtering results of the plurality of features and the single feature to determine the target threshold corresponding to the target parameter, which can more accurately determine the target threshold corresponding to the target parameter, so as to help improve the model effect of the service rule model when the target threshold corresponding to the target parameter is applied to the service rule model, better assist business decision-making, effectively reduce the resource occupation of the system where the service rule model is located, and improve work efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0017] The accompanying drawings, which are incorporated herein and form a part of the specification, illustrate embodiments consistent with the present application and, together with the description, further serve to explain the principles of the application.

[0018] Figure 1 An application scenario diagram is provided for an embodiment of the present application.

[0019] Figure 2 A flowchart of a threshold determination method is provided for an embodiment of the present application.

[0020] Figure 3 A processing flowchart of an unsupervised rule model is provided for an embodiment of the present application.

[0021] Figure 4 A structural diagram of a threshold determination device is provided for an embodiment of the present application.

[0022] Figure 5 A structural diagram of an electronic device is provided for an embodiment of the present application.

[0023] The specific embodiments of the application will be described in detail below with reference to the drawings. These drawings and the associated description are not intended to limit the scope of the application in any way, but merely to illustrate the concept of the application to those skilled in the art by reference to specific embodiments. DETAILED DESCRIPTION

[0024] The exemplary embodiments will be described in detail herein with reference to the attached drawings. The same reference numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments are not meant to represent all embodiments consistent with the application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the application as detailed in the appended claims.

[0025] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of related data comply with relevant laws, regulations and standards, necessary security measures are taken, do not violate public order and good customs, and provide corresponding operation portal for user to choose authorization or refusal.

[0026] And the present application involves big data analysis of user information (including but not limited to personal biological characteristics, identity data, consumption data, asset data, electronic terminal operation data, etc.), and uses artificial intelligence technology for automatic decision making, and makes technical solutions based on automatic decision making results that have a significant impact on personal rights and interests, provides corresponding operation portal for users to choose to agree or refuse automatic decision making results; if the user chooses to refuse, the expert decision making process is entered.

[0027] It should be noted that the threshold determination method, device, equipment, storage medium and program product provided by the present application can be used in the field of artificial intelligence, and can also be used in any field other than artificial intelligence. The application field of the threshold determination method, device, equipment, storage medium and program product in the present application is not limited.

[0028] In the field of marketing, operation and risk control, business rule model is needed to evaluate the effect of decision making according to the result of business rule model, and to adjust and optimize the management strategy in time.

[0029] Currently, thresholds for business rule models are typically determined based on expert experience. However, this method lacks analysis of actual business data, leading to inaccurate threshold settings. This results in unreasonable thresholds that fail to meet decision-making needs, causing multiple reworks in the system (e.g., a marketing system), wasting resources, and impacting work efficiency.

[0030] To address the aforementioned issues, this application provides a threshold determination method. Based on actual business data, an outlier screening method is applied to an unsupervised rule model. The method integrates the outlier screening results of multiple features and a single feature obtained from the business data to determine the target threshold corresponding to the target parameter to be determined. This method can more accurately determine the target threshold corresponding to the target parameter. Therefore, when the target threshold corresponding to the target parameter is applied to the business rule model, it helps to improve the model performance of the business rule model, better assists business decision-making, effectively reduces the resource consumption of the system where the business rule model is located, and improves work efficiency.

[0031] The following section provides examples illustrating the application scenarios of the solution provided in this application.

[0032] Figure 1 This is a schematic diagram illustrating an application scenario provided by an embodiment of this application. For example... Figure 1 As shown, in this application scenario, for each of the multiple business rule models (i.e., business rule model 1 to business rule model n), multiple business data samples for the target business rule are input into the unsupervised rule model on server 101. The unsupervised rule model determines the threshold corresponding to the parameter to be determined in the business rule model, and then applies the threshold corresponding to the parameter to the business rule model.

[0033] It should be noted that, Figure 1 This is merely a schematic diagram illustrating one application scenario provided by an embodiment of this application. This embodiment does not necessarily represent... Figure 1 The included equipment is not limited, nor is it restricted. Figure 1 The positional relationships between the devices are defined.

[0034] 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 be described below with reference to the accompanying drawings.

[0035] Figure 2A flowchart of a threshold determination method provided by an embodiment of the present application is shown. The threshold determination method can be executed by software and / or a hardware device, for example, the hardware device can be a threshold determination device, which can be an electronic device or a processing chip in an electronic device. As shown in Figure 2 The method of the embodiment of the present application includes the following steps.

[0036] S201: Obtain a plurality of business data samples for a target business rule, the target business rule including a business rule application target, a definition of the business rule, a target parameter for which a threshold value is to be determined, and customer attribute information.

[0037] In the embodiment of the present application, the target business rule is the business rule information required for constructing an unsupervised rule model, and the target business rule includes a business rule application target, a definition of the business rule, a target parameter for which a threshold value is to be determined, and customer attribute information. For example, the business rule application target is to identify abnormal transactions of elderly users when purchasing funds; the definition of the business rule is that a user whose age is above a certain age and whose risk level is above a certain level makes a purchase of a high-risk fund as an abnormal transaction, and according to the business rule application target and the definition of the business rule, the target parameter for which a threshold value is to be determined is the customer age and the customer risk level; and the customer attribute information includes, for example, the age, risk level, income, assets, and investment period of a specific customer.

[0038] In this step, a plurality of business data samples for a target business rule can be obtained, and the plurality of business data samples can also be understood as a plurality of historical business data samples, and the plurality of business data samples are used to construct an unsupervised rule model. The business data sample contains attribute values corresponding to customer attribute information, for example, the business data sample includes data values corresponding to customer attribute information such as the age, risk level, income, assets, and investment period of a specific customer.

[0039] S202: Input the plurality of business data samples into an unsupervised rule model, the unsupervised rule model performing feature extraction on the business data samples to obtain a first feature vector corresponding to a plurality of features and a second feature vector corresponding to a single feature, the plurality of features being determined according to the business data samples, and the single feature being determined according to the target parameter; filter the business data samples based on the first feature vector and the second feature vector to obtain first abnormal business data samples corresponding to the first feature vector and second abnormal business data samples corresponding to the second feature vector; and determine a target threshold value corresponding to the target parameter according to the first abnormal business data samples and the second abnormal business data samples, the target threshold value being used as a threshold value of the target parameter in a business rule model.

[0040] It can be understood that unsupervised mainly refers to that, in a modeling process, a model only accepts a feature vector of a sample as input without using defined label information, and the model needs to mine potential information, structure or rule inside data by itself. For example, "monitoring customer's behavior of purchasing high-risk funds" is a method for identifying abnormal transactions in business, but not all behaviors of purchasing high-risk funds are abnormal, that is, the purchase or specific purchase times cannot be used as model labels to distinguish good and bad samples. At this time, the business rule model is unsupervised.

[0041] In this step, the plurality of business data samples are input into the unsupervised rule model, and the unsupervised rule model extracts features from the business data samples to obtain a first feature vector corresponding to a plurality of features and a second feature vector corresponding to a single feature. The plurality of features are determined according to the business data samples. For example, the customer attribute information contained in the business data samples includes the age, risk level, income, asset and investment period of a specific customer, and then five features corresponding to the age, risk level, income, asset and investment period can be obtained.

[0042] Optionally, a plurality of features with greater relevance to the target parameter can also be selected from the business data samples by using a preset feature selection method, which can include at least one of a correlation analysis (Pearson correlation coefficient) method, a clustering method and a machine learning model (tree model). The single feature is determined according to the target parameter. For example, the target parameter is the age of a customer, and then the single feature can be the age of the customer.

[0043] Optionally, inputting the plurality of business data samples into the unsupervised rule model and extracting features from the business data samples to obtain the first feature vector corresponding to the plurality of features and the second feature vector corresponding to the single feature can include: preprocessing the plurality of business data samples to obtain preprocessed samples, the preprocessing including at least one of data standardization processing, data outlier processing and data missing value filling processing; and extracting features from the preprocessed samples by the unsupervised rule model to obtain the first feature vector corresponding to the plurality of features and the second feature vector corresponding to the single feature.

[0044] For example, to ensure the data quality of the business data samples, the business data samples can be preprocessed to obtain preprocessed samples. For example, the business data samples can be subjected to data standardization processing, data outlier processing according to a preset numerical range, and data missing value filling processing. The unsupervised rule model extracts features from the preprocessed samples to obtain the first feature vector corresponding to the plurality of features and the second feature vector corresponding to the single feature.

[0045] After obtaining the first feature vectors corresponding to the plurality of features and the second feature vector corresponding to the single feature, the business data samples can be screened based on the first feature vectors and the second feature vector to obtain first abnormal business data samples corresponding to the first feature vectors and second abnormal business data samples corresponding to the second feature vector. For example, based on the first feature vectors and the second feature vector, a statistical method or a machine learning method can be used to screen the business data samples to obtain the first abnormal business data samples corresponding to the first feature vectors and the second abnormal business data samples corresponding to the second feature vector. For specific ways of obtaining the first abnormal business data samples and the second abnormal business data samples, reference can be made to subsequent embodiments.

[0046] It can be understood that although the unsupervised rule model in the unsupervised case does not have a clear definition of positive samples and negative samples, the abnormal business data samples can be screened by a statistical method or a machine learning method, so that the abnormal boundary values can be distinguished by using the abnormal business data samples, and then the threshold required by the unsupervised rule model can be confirmed. When selecting a specific screening method, the characteristics of the business data, the requirements of the problem, and the limitations of the computing resources can be considered comprehensively. Generally, the statistical method has stronger business interpretability, and the machine learning method can more accurately identify the abnormal business data samples.

[0047] After obtaining the first abnormal business data samples and the second abnormal business data samples, the abnormal boundary value corresponding to the target parameter can be determined according to the first abnormal business data samples and the second abnormal business data samples. The abnormal boundary value is the target threshold corresponding to the target parameter. For example, the first threshold corresponding to the target parameter can be determined according to the first abnormal business data samples, and the second threshold corresponding to the target parameter can be determined according to the second abnormal business data samples. Then, the first threshold and the second threshold are dynamically fused to determine the abnormal boundary value corresponding to the target parameter, and the abnormal boundary value is taken as the target threshold corresponding to the target parameter. Referring to the example in the S201 step, the target parameters to be determined threshold are, for example, customer age and customer risk level, and the target threshold corresponding to the customer age and the target threshold corresponding to the customer risk level can be determined by the unsupervised rule model, respectively.

[0048] After determining the target threshold corresponding to the target parameter, the target threshold corresponding to the target parameter can be applied to the business rule model.

[0049] Optionally, after determining the target threshold corresponding to the target parameter, the threshold determination method provided by the embodiments of the present application can further include: determining whether to correct the target threshold based on a preset threshold range; if yes, correcting the target threshold to obtain a corrected threshold, and the corrected threshold is used as the threshold of the target parameter in the business rule model.

[0050] In consideration of the fact that the target threshold corresponding to the target parameter determined by the unsupervised rule model may not conform to the business logic, such as being much higher or much lower than the business expectation, the target threshold can be corrected based on a preset threshold range. By way of example, the preset threshold range is a threshold range predefined based on business experience. The target parameter is, for example, customer age, and the preset threshold range is, for example, 54 to 56 years old. Assuming that the target threshold corresponding to the target parameter determined by the unsupervised rule model is 55 years old, which is within the preset threshold range, it can be determined that the target threshold does not need to be corrected. If the target threshold corresponding to the target parameter determined by the unsupervised rule model is 80 years old, which is outside the preset threshold range, it is determined that the target threshold needs to be corrected, and the target threshold can be corrected based on the preset threshold range, such as taking the median of the preset threshold range as the corrected threshold.

[0051] The threshold determination method provided by the embodiments of the present application comprises the following steps: obtaining a plurality of business data samples for a target business rule, the target business rule comprising a business rule application target, a definition of the business rule, a target parameter for which a threshold is to be determined, and customer attribute information; inputting the plurality of business data samples into an unsupervised rule model, the unsupervised rule model performing feature extraction on the business data samples to obtain a first feature vector corresponding to a plurality of features and a second feature vector corresponding to a single feature, the plurality of features being determined according to the business data samples, and the single feature being determined according to the target parameter; filtering the business data samples based on the first feature vector and the second feature vector to obtain first abnormal business data samples corresponding to the first feature vector and second abnormal business data samples corresponding to the second feature vector; and determining a target threshold corresponding to the target parameter according to the first abnormal business data samples and the second abnormal business data samples, the target threshold being used as a threshold of the target parameter in a business rule model. In the embodiments of the present application, the first abnormal business data samples corresponding to the first feature vector and the second abnormal business data samples corresponding to the second feature vector are obtained by filtering using the unsupervised rule model, and the target threshold corresponding to the target parameter is determined according to the first abnormal business data samples and the second abnormal business data samples. The abnormal value filtering method is used in the unsupervised rule model based on actual business data conditions, and the abnormal value filtering results of the plurality of features and the single feature are combined to determine the target threshold corresponding to the target parameter. The target threshold corresponding to the target parameter can be determined more accurately, which helps to improve the model effect of the business rule model when the target threshold corresponding to the target parameter is applied to the business rule model, better assists business decision-making, effectively reduces the resource occupation of the system in which the business rule model is located, and improves work efficiency.

[0052] On the basis of the above embodiments, Figure 3 The processing flowchart of the unsupervised rule model provided by an embodiment of the present application is shown in FIG. 2. As shown in FIG. 2, the processing flowchart comprises the following steps:Figure 3 As shown, the processing flow of the unsupervised rule model of the embodiments of the present application can include:

[0053] S301, feature extraction is performed on the input service data sample to obtain a first feature vector corresponding to a plurality of features and a second feature vector corresponding to a single feature, the plurality of features being determined according to the service data sample, and the single feature being determined according to the target parameter.

[0054] The specific description of this step can be referred to Figure 2 The related description of S202 in the embodiments is not repeated here.

[0055] S302, based on the first feature vector and the second feature vector, a preset screening method is used to screen the service data sample to obtain a first abnormal service data sample corresponding to the first feature vector and a second abnormal service data sample corresponding to the second feature vector, and the preset screening method includes at least one of a statistical-based method, a clustering-based method and a tree-based method.

[0056] It can be understood that the statistical-based method is usually concerned with inferring and explaining the data generation process, emphasizes the interpretability and hypothesis testing of the model, and is commonly used to understand the relationship between variables. Optionally, the statistical-based method can include a 3Sigma criterion method and / or an Interquartile Range (IQR) method.

[0057] Exemplarily, the 3Sigma criterion method is based on the characteristics of normal distribution to evaluate the distribution rule and probability range of data; when the sample data is identified as a normal distribution, the abnormal value is located by the 3Sigma principle, and when the data is distributed within 3 standard deviations around the mean value, it is determined that the data point is abnormal, at this time, the 3 standard deviations around the mean value is the abnormal value boundary, that is, the possible threshold range. The Interquartile Range method is a measure used in statistics to describe the width of data distribution or the degree of variation, which is equal to the difference between the third quartile (Q3, i.e. 75% quantile) and the first quartile (Q1, i.e. 25% quantile); when identifying abnormal values, the upper and lower boundaries are determined by Q1 and Q3 respectively, and the data points outside the boundary within a certain IQR multiple (such as 1.5 times IQR) range are usually determined as abnormal values, and the boundary at this time is the abnormal value boundary.

[0058] The clustering-based method and the tree-based method are both machine learning methods. Optionally, the clustering-based method can include a Density-Based Spatial Clustering of Applications with Noise (DBSCAN) method.

[0059] Exemplarily, the clustering-based method groups data points into clusters with similar characteristics, while outliers are usually represented as points that are not closely related to any cluster. The DBSCAN method can identify clusters of arbitrary shape and regard points not belonging to any cluster as outliers. The DBSCAN algorithm defines two important parameters, i.e., a neighborhood radius (ε) and a threshold of the number of data objects in the neighborhood (MinPts). If an ε-neighborhood of a data object contains at least MinPts data objects, the data object is determined as a core point; if a data object is not a core point but is located in the ε-neighborhood of a core point, the data object is determined as a boundary point; other data objects that are neither core points nor boundary points are determined as noise points, i.e., outliers, and the abnormal boundary value can be determined by analyzing the feature values corresponding to the abnormal business data samples.

[0060] Optionally, the tree-based method can include an isolated forest (iForest) method. Exemplarily, the isolated forest method is suitable for anomaly detection of continuous data, which defines an anomaly as an "easily isolated outlier", i.e., a point that is sparsely distributed and far from a high-density population. The isolated forest method mainly includes two steps: first, training iForest: sampling from the training set, constructing isolated trees, testing each isolated tree in the forest, and recording the path length; second, calculating the anomaly score: according to the anomaly score calculation formula, the anomaly score of each sample point is obtained; points with anomaly scores close to 1 are determined as abnormal points, while points with anomaly scores much less than 0.5 are determined as normal points. The abnormal boundary value can be determined by analyzing the feature values corresponding to the abnormal business data samples.

[0061] In this step, exemplarily, the business data samples can be screened based on the first feature vector and the second feature vector using a preset screening method to obtain first abnormal business data samples corresponding to the first feature vector and second abnormal business data samples corresponding to the second feature vector. For example, the screening results of one or more preset screening methods can be combined to determine the abnormal boundary value. It can be understood that when selecting a specific screening method, the characteristics of the business data, the requirements of the problem, and the limitations of the computing resources can be considered comprehensively. Generally, the statistical-based method has stronger business interpretability, while the machine learning method can more accurately identify abnormal business data samples.

[0062] S303, determining a first threshold value corresponding to the target parameter according to the parameter value of the target parameter in the first abnormal business data sample.

[0063] In this step, after obtaining the first abnormal business data sample corresponding to the first feature vector, the first threshold value corresponding to the target parameter can be determined according to the parameter value of the target parameter in the first abnormal business data sample.

[0064] Further, optionally, determining the first threshold value corresponding to the target parameter according to the parameter value of the target parameter in the first abnormal service data sample can include: determining the first threshold value corresponding to the target parameter as an average value of the parameter value of the target parameter in the first abnormal service data sample.

[0065] For example, the target parameter is the customer age, and the average value of the customer age in the first abnormal service data sample can be obtained as the first threshold value corresponding to the target parameter.

[0066] S304, determining the second threshold value corresponding to the target parameter according to the parameter value of the target parameter in the second abnormal service data sample.

[0067] In this step, after obtaining the second abnormal service data sample corresponding to the second feature vector, the second threshold value corresponding to the target parameter can be determined according to the parameter value of the target parameter in the second abnormal service data sample.

[0068] Further, optionally, determining the second threshold value corresponding to the target parameter according to the parameter value of the target parameter in the second abnormal service data sample can include: determining the second threshold value corresponding to the target parameter as an average value of the parameter value of the target parameter in the second abnormal service data sample.

[0069] For example, the target parameter is the customer age, and the average value of the customer age in the second abnormal service data sample can be obtained as the second threshold value corresponding to the target parameter.

[0070] It should be noted that the order of the steps S303 and S304 is not limited by the embodiments of the present application.

[0071] S305, weighting and fusing the first threshold value and the second threshold value to determine the target threshold value corresponding to the target parameter.

[0072] In this step, after obtaining the first threshold value and the second threshold value corresponding to the target parameter, the first threshold value and the second threshold value can be weighted and summed according to the first threshold value, the first weight corresponding to the first threshold value, the second threshold value and the second weight corresponding to the second threshold value, to obtain the target threshold value corresponding to the target parameter.

[0073] After obtaining the target threshold value corresponding to the target parameter, the target threshold value corresponding to the target parameter can be applied to the service rule model.

[0074] In the embodiment of the present application, the input business data sample is subjected to feature extraction to obtain a first feature vector corresponding to a plurality of features and a second feature vector corresponding to a single feature, the plurality of features being determined according to the business data sample, and the single feature being determined according to the target parameter; based on the first feature vector and the second feature vector, a preset screening method is used to screen the business data sample to obtain a first abnormal business data sample corresponding to the first feature vector and a second abnormal business data sample corresponding to the second feature vector, the preset screening method including at least one of a statistical method, a clustering method and a tree-based method, which can ensure the screening effect of the abnormal business data sample; a first threshold value corresponding to the target parameter is determined according to the parameter value of the target parameter in the first abnormal business data sample, a second threshold value corresponding to the target parameter is determined according to the parameter value of the target parameter in the second abnormal business data sample, the first threshold value and the second threshold value are weighted and fused to determine a target threshold value corresponding to the target parameter, and the target threshold value is used as a threshold value of the target parameter in the business rule model, so as to capture a complex correlation mode through the first threshold value corresponding to the plurality of features, provide explainability bottom line through the second threshold value corresponding to the single feature, and fuse the first threshold value and the second threshold value to determine the target threshold value corresponding to the target parameter, thereby more accurately determining the target threshold value corresponding to the target parameter.

[0075] Based on the above embodiment, in an actual business scenario, for example, the core target of a marketing system is to accurately reach target users and improve conversion rate and return on investment, the threshold value of the business rule model determined by the threshold value determination method provided by the present application can be used to more accurately determine whether a user belongs to a target group, whether to issue a discount or trigger a marketing activity. The operation system relies on business rules for user stratification, benefit allocation, activity triggering, etc., and the threshold value of the business rule model determined by the threshold value determination method provided by the present application can improve operation efficiency and user experience. The core of the risk control system is to identify and avoid risks, and the threshold value of the business rule model determined by the threshold value determination method provided by the present application can improve the accuracy of risk identification. In summary, the threshold value of the business rule model determined by the threshold value determination method provided by the present application can better improve the model effect of the business rule model, better assist business decision-making, effectively reduce the resource occupation of the system where the business rule model is located, and improve work efficiency.

[0076] The following is an embodiment of the device of the present application, which can be used to execute the method embodiments of the present application. For details not disclosed in the device embodiments of the present application, please refer to the method embodiments of the present application.

[0077] Figure 4 The structure diagram of the threshold value determination device provided by an embodiment of the present application is shown in FIG. 4. As shown in FIG. 4, the threshold value determination device 400 of the embodiment of the present application includes an acquisition module 401 and a processing module 402. Figure 4 The acquisition module 401 is configured to acquire a business data sample, and the processing module 402 is configured to perform feature extraction on the business data sample to obtain a first feature vector corresponding to a plurality of features and a second feature vector corresponding to a single feature, the plurality of features being determined according to the business data sample, and the single feature being determined according to a target parameter; based on the first feature vector and the second feature vector, a preset screening method is used to screen the business data sample to obtain a first abnormal business data sample corresponding to the first feature vector and a second abnormal business data sample corresponding to the second feature vector, the preset screening method including at least one of a statistical method, a clustering method and a tree-based method, which can ensure the screening effect of the abnormal business data sample; a first threshold value corresponding to the target parameter is determined according to the parameter value of the target parameter in the first abnormal business data sample, a second threshold value corresponding to the target parameter is determined according to the parameter value of the target parameter in the second abnormal business data sample, the first threshold value and the second threshold value are weighted and fused to determine a target threshold value corresponding to the target parameter, and the target threshold value is used as a threshold value of the target parameter in the business rule model, so as to capture a complex correlation mode through the first threshold value corresponding to the plurality of features, provide explainability bottom line through the second threshold value corresponding to the single feature, and fuse the first threshold value and the second threshold value to determine the target threshold value corresponding to the target parameter, thereby more accurately determining the target threshold value corresponding to the target parameter.

[0078] The acquisition module 401 is configured to acquire a plurality of service data samples for a target service rule, the target service rule including a service rule application target, a definition of the service rule, a target parameter to be determined, and customer attribute information.

[0079] The processing module 402 is configured to input the plurality of service data samples into an unsupervised rule model, the unsupervised rule model performing feature extraction on the service data samples to obtain a first feature vector corresponding to a plurality of features and a second feature vector corresponding to a single feature, the plurality of features being determined according to the service data samples, and the single feature being determined according to the target parameter; performing screening on the service data samples based on the first feature vector and the second feature vector to obtain first abnormal service data samples corresponding to the first feature vector and second abnormal service data samples corresponding to the second feature vector; and determining a target threshold corresponding to the target parameter according to the first abnormal service data samples and the second abnormal service data samples, the target threshold being used as a threshold of the target parameter in a service rule model.

[0080] In some embodiments, when the processing module 402 is configured to determine the target threshold corresponding to the target parameter according to the first abnormal service data samples and the second abnormal service data samples, the processing module 402 can be specifically configured to: determine a first threshold corresponding to the target parameter according to a parameter value of the target parameter in the first abnormal service data samples; determine a second threshold corresponding to the target parameter according to a parameter value of the target parameter in the second abnormal service data samples; and perform weighted fusion on the first threshold and the second threshold to determine the target threshold corresponding to the target parameter.

[0081] Optionally, when the processing module 402 is configured to determine the first threshold corresponding to the target parameter according to the parameter value of the target parameter in the first abnormal service data samples, the processing module 402 can be specifically configured to: determine the first threshold corresponding to the target parameter as an average value of the parameter value of the target parameter in the first abnormal service data samples.

[0082] Optionally, when the processing module 402 is configured to determine the second threshold corresponding to the target parameter according to the parameter value of the target parameter in the second abnormal service data samples, the processing module 402 can be specifically configured to: determine the second threshold corresponding to the target parameter as an average value of the parameter value of the target parameter in the second abnormal service data samples.

[0083] In some embodiments, when the processing module 402 filters business data samples based on a first feature vector and a second feature vector to obtain a first abnormal business data sample corresponding to the first feature vector and a second abnormal business data sample corresponding to the second feature vector, it may specifically be used to: filter business data samples based on the first feature vector and the second feature vector using a preset filtering method to obtain a first abnormal business data sample corresponding to the first feature vector and a second abnormal business data sample corresponding to the second feature vector. The preset filtering method includes at least one of a statistical method, a clustering method, and a tree-based method.

[0084] Optionally, when the processing module 402 is used to input multiple business data samples into an unsupervised rule model, and the unsupervised rule model extracts features from the business data samples to obtain a first feature vector corresponding to multiple features and a second feature vector corresponding to a single feature, the specific processing can be as follows: preprocessing the multiple business data samples to obtain preprocessed samples, wherein the preprocessing includes at least one of data standardization, outlier handling, and missing value imputation; and the unsupervised rule model extracts features from the preprocessed samples to obtain a first feature vector corresponding to multiple features and a second feature vector corresponding to a single feature.

[0085] Optionally, the processing module 402 can also be used to: after determining the target threshold corresponding to the target parameter, determine whether to correct the target threshold based on a preset threshold range; if so, correct the target threshold to obtain the corrected threshold, and use the corrected threshold as the threshold of the target parameter in the business rule model.

[0086] The threshold determination device of this application embodiment can be used to execute the technical solution of any of the method embodiments shown above. Its implementation principle and technical effect are similar, and will not be repeated here.

[0087] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 5 As shown, the electronic device 500 may include at least one processor 501 and a memory 502.

[0088] Memory 502 is used to store programs. Specifically, the program may include program code, which includes computer-executable instructions.

[0089] The memory 502 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device.

[0090] The processor 501 executes computer execution instructions stored in the memory 502 to implement the threshold determination method described in the foregoing method embodiments. The processor 501 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this application. Specifically, when implementing the threshold determination method described in the foregoing method embodiments, the electronic device may be, for example, a server or other electronic device with processing capabilities.

[0091] Optionally, the electronic device 500 may also include a communication interface 503. In specific implementations, if the communication interface 503, memory 502, and processor 501 are implemented independently, they can be interconnected via a bus to complete communication. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc., but this does not imply that there is only one bus or one type of bus.

[0092] Optionally, in a specific implementation, if the communication interface 503, memory 502, and processor 501 are integrated on a single chip, then the communication interface 503, memory 502, and processor 501 can communicate through an internal interface.

[0093] The electronic device is used to execute the threshold determination method in any of the foregoing method embodiments. Its implementation principle and technical effect are similar, and will not be described again here.

[0094] This application also provides a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implements the above-described threshold determination method.

[0095] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described threshold determination method.

[0096] The aforementioned computer-readable storage medium can be implemented from 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. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0097] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an application-specific integrated circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in a threshold determination device.

[0098] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0099] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0100] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0101] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory, random access memory (RAM), magnetic disks, or optical disks.

[0102] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0103] Finally, it should be noted that other embodiments of this application will readily conceive of by 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, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and alterations may be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A threshold determination method, characterized in that, include: Obtain multiple business data samples for the target business rule, wherein the target business rule includes the application target of the business rule, the definition of the business rule, the target parameters of the threshold to be determined, and customer attribute information; The multiple business data samples are input into an unsupervised rule model, which extracts features from the business data samples to obtain a first feature vector corresponding to multiple features and a second feature vector corresponding to a single feature. The multiple features are determined based on the business data samples, and the single feature is determined based on the target parameter. The business data samples are then filtered based on the first feature vector and the second feature vector to obtain a first abnormal business data sample corresponding to the first feature vector and a second abnormal business data sample corresponding to the second feature vector. Based on the first abnormal business data sample and the second abnormal business data sample, a target threshold corresponding to the target parameter is determined, and the target threshold is used as the threshold of the target parameter in the business rule model.

2. The threshold determination method according to claim 1, characterized in that, The step of determining the target threshold corresponding to the target parameter based on the first abnormal business data sample and the second abnormal business data sample includes: Based on the parameter value of the target parameter in the first abnormal business data sample, determine the first threshold corresponding to the target parameter; Based on the parameter value of the target parameter in the second abnormal business data sample, determine the second threshold corresponding to the target parameter; The first threshold and the second threshold are weighted and fused to determine the target threshold corresponding to the target parameter.

3. The threshold determination method according to claim 2, characterized in that, The step of determining the first threshold corresponding to the target parameter based on the parameter value of the target parameter in the first abnormal business data sample includes: The first threshold corresponding to the target parameter is determined to be the average value of the target parameter in the first abnormal business data sample.

4. The threshold determination method according to claim 2, characterized in that, The step of determining the second threshold corresponding to the target parameter based on the parameter value of the target parameter in the second abnormal business data sample includes: The second threshold corresponding to the target parameter is determined to be the average value of the target parameter in the second abnormal business data sample.

5. The threshold determination method according to claim 1, characterized in that, The step of filtering the business data samples based on the first feature vector and the second feature vector to obtain a first abnormal business data sample corresponding to the first feature vector and a second abnormal business data sample corresponding to the second feature vector includes: Based on the first feature vector and the second feature vector, a preset filtering method is used to filter the business data samples to obtain a first abnormal business data sample corresponding to the first feature vector and a second abnormal business data sample corresponding to the second feature vector. The preset filtering method includes at least one of statistical methods, clustering methods, and tree-based methods.

6. The threshold determination method according to any one of claims 1 to 5, characterized in that, The step of inputting the multiple business data samples into an unsupervised rule model, wherein the unsupervised rule model performs feature extraction on the business data samples to obtain a first feature vector corresponding to multiple features and a second feature vector corresponding to a single feature, includes: The multiple business data samples are preprocessed to obtain preprocessed samples. The preprocessing includes at least one of data standardization, outlier handling, and missing value imputation. The unsupervised rule model extracts features from the preprocessed samples to obtain a first feature vector corresponding to the multiple features and a second feature vector corresponding to the single feature.

7. The threshold determination method according to any one of claims 1 to 5, characterized in that, After determining the target threshold corresponding to the target parameter, the method further includes: Based on a preset threshold range, determine whether to correct the target threshold; If so, the target threshold is corrected to obtain a corrected threshold, which is used as the threshold of the target parameter in the business rule model.

8. A threshold determination device, characterized in that, include: The acquisition module is used to acquire multiple business data samples for the target business rule, wherein the target business rule includes the application target of the business rule, the definition of the business rule, the target parameters of the threshold to be determined, and customer attribute information; The processing module is configured to input the multiple business data samples into an unsupervised rule model, wherein the unsupervised rule model extracts features from the business data samples to obtain a first feature vector corresponding to multiple features and a second feature vector corresponding to a single feature. The multiple features are determined based on the business data samples, and the single feature is determined based on the target parameter. The processing module then filters the business data samples based on the first and second feature vectors to obtain a first abnormal business data sample corresponding to the first feature vector and a second abnormal business data sample corresponding to the second feature vector. Finally, based on the first and second abnormal business data samples, a target threshold corresponding to the target parameter is determined, and the target threshold is used as the threshold of the target parameter in the business rule model.

9. 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 threshold determination method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions, which, when executed, implement the threshold determination method as described in any one of claims 1 to 7.

11. A computer program product, comprising a computer program, characterized in that, When the computer program is executed, it implements the threshold determination method as described in any one of claims 1 to 7.

Images