Business rule compliance customer quantity prediction method, device, equipment, medium and product

Abstract

The present application relates to the technical field of data processing, and particularly relates to a business rule compliance customer quantity prediction method, device, equipment, medium and product, the method automatically identifies business association parameters and performs classification processing by analyzing user-configured business rule information, calculates and generates a standardized prediction template vector based on the parameter type and the original configuration rule, inputs the vector into a pre-trained prediction model to obtain an estimated number of compliance customers. The present application automatically analyzes and converts unstructured business rules into a standardized vector representation, so that business rules of different types and complexities can be efficiently processed under the same model framework, thereby improving the standardization and efficiency of the prediction process. In addition, the introduced degradation processing mechanism can provide an alternative prediction scheme when the model confidence is low, avoiding the risk of a single model failure and enhancing the robustness and usability of the entire prediction system in practical applications.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to methods, apparatus, equipment, media and products for predicting the number of customers who meet business rules. Background Technology

[0002] As marketing services undergo a deeper digital transformation, platforms need to accurately reach target customers through various marketing activities, and customer qualification assessment platforms have become the core system supporting this type of business. These platforms, by pre-setting a large number of business rule templates, enable multi-dimensional screening and judgment of customer attributes to meet the qualification verification needs of different marketing scenarios. Due to the increasing complexity and personalization of business scenarios, the number of rule templates accumulated on the platform continues to grow, and their parameter combinations are becoming increasingly diverse.

[0003] In actual business operations, after configuring rule parameters, business personnel often need to estimate the number of customers who meet the rule in advance in order to rationally plan equity resources and control activity costs. However, the current implementation mainly adopts the method of manually writing and executing query scripts in the data warehouse for analysis. The whole process relies on manual intervention and requires cross-system operation, resulting in a significant time delay between configuration completion and obtaining the estimated result. While some methods based on off-warehouse marketplaces can avoid resource contention in online services to some extent, they still rely on full table scans and batch calculations of massive amounts of intermediate data. This not only consumes a lot of storage and computing resources, but also causes the estimation time to be as long as several minutes, which cannot meet the agile needs of business personnel for real-time preview and rapid parameter adjustment.

[0004] Therefore, how to quickly and efficiently estimate the number of qualified customers after configuring business rules, and reduce system resource consumption, has become a technical problem that urgently needs to be solved in this industry. Summary of the Invention

[0005] The main objective of this invention is to provide a method, apparatus, device, medium, and product for predicting the number of customers who meet business rules, aiming to solve the technical problem in the prior art of how to quickly and efficiently estimate the number of customers who meet the criteria after the business rules are configured, and reduce the consumption of system resources.

[0006] To achieve the above objectives, the present invention provides a method for predicting the number of customers meeting business rules, the method comprising the following steps:

[0007] Receive user-configured business rule information; Perform semantic analysis on the business rule information to determine business-related parameters; The business-related parameters are classified according to variable type to obtain the classification results of the business-related parameters; Based on the classification results of the business association parameters and the original data in the business rule information, calculate the feature values ​​of the business association parameters in the template vector; Based on the feature values ​​of the business-related parameters, a prediction template vector is constructed; The prediction template vector is input into the customer number prediction model to obtain the predicted number of qualified customers based on the business rule information.

[0008] Optionally, the step of performing semantic analysis on the business rule information to determine business association parameters includes: Based on the business rule information, and combined with a predefined business domain dictionary, basic semantic analysis is performed to determine key business terms; Based on the key business terms, semantic mapping is performed with reference to the parameter definition knowledge base to determine their corresponding standardized parameter names; The standardized parameter name is combined with the corresponding value in the business rule information to obtain parameter value pairs; Logical consistency verification is performed on the parameter value pairs to obtain the business-related parameters.

[0009] Optionally, calculating the feature value of the business association parameter in the template vector based on the classification result of the business association parameter and the original data in the business rule information includes: When the classification result of the business-related parameter is an enumerated parameter, the original data of the business-related parameter in the business rule information is obtained; Based on the original data of the business-related parameters, determine the selected enumeration items contained in the original data; Based on the enumeration item business weight record table corresponding to the business association parameters, determine the original weight value of the selected enumeration item; Calculate the weight feature value of the business association parameter based on the original weight value of the selected enumeration item; Based on the weighted feature values, the feature values ​​of the business association parameters in the template vector are obtained.

[0010] Optionally, calculating the weight feature value of the business association parameter based on the original weight value of the selected enumeration item includes: Based on the original weight values ​​of each selected enumeration item, determine the total weight and the extreme value of the weight; Based on the sum of the weights, calculate the average weight and standard deviation of the weights formed by the selected enumeration items; The sum of weights, extreme values ​​of weights, average values ​​of weights, and standard deviation of weights formed by the selected enumeration items included in the business association parameters are used as the weight feature values ​​of the business association parameters.

[0011] Optionally, constructing a prediction template vector based on the feature values ​​of the business association parameters includes: Based on a preset parameter location index table, determine the fixed writing position of the business-related parameter in the prediction template vector; Based on the classification type of the business association parameters, determine the number of fixed dimensions of the feature values ​​in the prediction template vector; The feature values ​​of the business-related parameters are standardized according to a preset data format; The standardized feature values ​​are filled into a fixed-length vector frame according to the fixed writing position and fixed number of dimensions to obtain the prediction template vector.

[0012] Optionally, the method for predicting the number of customers meeting the business rules further includes: When the confidence level of the predicted number of qualified customers output by the customer number prediction model is lower than a preset threshold, the downgrade process is initiated. The downgrade process includes: Obtain the preset sampling ratio configuration information; Based on the sampling ratio configuration information, sample data is extracted from the intermediate data table associated with the business rules to construct a sample data table; The number of sample customers that meet the business rules is used as the number of samples that meet the criteria, based on the sample data table. Based on the number of samples that meet the standards, the total number of customers, and the total amount of sample data, the estimated number of customers that meet the standards is determined, and the estimated number of customers that meet the standards is used as the predicted number of customers that meet the standards.

[0013] Furthermore, to achieve the above objectives, the present invention also proposes a device for predicting the number of customers meeting business rule requirements, the device comprising: The business rule formulation module is used to receive business rule information configured by the user; The parameter parsing and classification module is used to perform semantic analysis on the business rule information and determine the business-related parameters; The parameter parsing and classification module is also used to classify the business-related parameters according to variable type to obtain the classification result of the business-related parameters; The feature value calculation module is used to calculate the feature value of the business association parameter in the template vector based on the classification result of the business association parameter and the original data in the business rule information. The model prediction module is used to construct a prediction template vector based on the feature values ​​of the business-related parameters; The model prediction module is also used to input the prediction template vector into the customer number prediction model to obtain the predicted number of qualified customers of the business rule information.

[0014] Furthermore, to achieve the above objectives, the present invention also proposes a business rule compliance customer number prediction device, which includes: a memory, a processor, and a business rule compliance customer number prediction program stored in the memory and executable on the processor. The business rule compliance customer number prediction program is configured to implement the steps of the business rule compliance customer number prediction method described above.

[0015] Furthermore, to achieve the above objectives, the present invention also proposes a storage medium storing a business rule compliance customer number prediction program, which, when executed by a processor, implements the steps of the business rule compliance customer number prediction method described above.

[0016] In addition, to achieve the above objectives, the present invention also proposes a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the business rule compliance customer number prediction method described above.

[0017] The proposed technical solutions (one or more) have at least the following technical effects: First, this solution receives user-configured business rule information and performs semantic analysis to intelligently identify and determine key business-related parameters. Then, it classifies these parameters based on their characteristics and converts the original business data into standardized feature values ​​based on the classification results. Using these feature values, a prediction template vector representing the core logic of the business rule is constructed. Finally, this vector is input into a pre-trained customer quantity prediction model, which directly outputs the predicted number of qualified customers under the business rule. Compared to existing solutions that require full table scans and batch calculations to obtain the qualified number, this solution abstracts complex business rules into lightweight feature vectors and uses a trained model for millisecond-level prediction and inference, reducing the waiting time from minutes or even days to seconds or even milliseconds. This allows business personnel to receive immediate feedback after configuring rules, greatly improving decision-making efficiency and user experience. In addition, the core of this solution lies in the prediction model and the feature extraction of the rules. There is no need to copy or process the huge original business data tables. Only a small number of model parameters and historical configuration features need to be stored, which fundamentally saves a lot of storage and computing resources. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the first embodiment of the method for predicting the number of customers meeting the business rules of the present invention. Figure 2 This is a flowchart illustrating the second embodiment of the method for predicting the number of customers meeting the business rules of the present invention. Figure 3 This is a flowchart illustrating the third embodiment of the method for predicting the number of customers meeting the business rules of the present invention. Figure 4 This is a structural block diagram of the first embodiment of the business rule compliance customer number prediction device of the present invention; Figure 5 This is a schematic diagram of the structure of the business rule compliance customer number prediction device in the hardware operating environment involved in the embodiments of the present invention.

[0021] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0022] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0023] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0024] Currently, due to increasingly complex and personalized business scenarios, after configuring rule parameters, business personnel primarily rely on manually writing and executing query scripts in the data warehouse for analysis. This process depends on manual intervention and requires cross-system operations, resulting in a significant time delay between configuration completion and obtaining the estimated results. While some methods based on off-warehouse marketplaces avoid resource contention in online services to some extent, they still fundamentally rely on full table scans and batch calculations of massive amounts of intermediate data. This not only consumes significant storage and computing resources but also leads to estimation times of several minutes, failing to meet the agile needs of business personnel for real-time preview and rapid parameter tuning. Therefore, how to quickly and efficiently estimate the number of qualified customers after configuring business rules, while reducing system resource consumption, is a pressing technical problem that needs to be solved.

[0025] This application receives user-configured business rule information and performs semantic analysis to intelligently identify and determine key business-related parameters. Subsequently, these parameters are classified based on their characteristics, and the raw business data is converted into standardized feature values ​​based on the classification results. Using these feature values, a prediction template vector representing the core logic of the business rule is constructed. Finally, this vector is input into a pre-trained customer quantity prediction model, which directly outputs the predicted number of qualified customers under the given business rule.

[0026] It should be noted that the executing entity of this invention can be a business rule compliance customer number prediction device, or a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or a thermal management device capable of realizing the above functions of a business rule compliance customer number prediction device, etc. This embodiment does not specifically limit it in this way. The following uses a business rule compliance customer number prediction device as the executing entity as an example to describe this embodiment and the following embodiments.

[0027] Based on this, this application provides a method for predicting the number of customers who meet business rules, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the method for predicting the number of customers meeting the business rules of this application.

[0028] In this embodiment, the method for predicting the number of customers meeting the business rules includes steps S10 to S60: Step S10: Receive the business rule information configured by the user.

[0029] It's important to note that in practical implementation, business rule information refers to the filtering criteria set by marketing campaign designers through a human-computer interaction interface. This process typically involves structured operations within a pre-designed rule configuration interface. For example, business personnel might select "VIP Customer" or "Regular Customer" from a dropdown menu, check "Account Opening City" as "Beijing," "Shanghai," and "Guangzhou" using checkboxes, and set the "Average Daily Account Balance" to be between 50,000 and 500,000 yuan using a slider or input box. Ultimately, the system integrates these discrete operation points into a complete, structured, machine-readable business rule definition. Its core purpose is to efficiently and unambiguously transform the user's business intent into data objects that the computer system can subsequently process.

[0030] Understandably, the business rule information received here is essentially a structured query instruction, defining a series of characteristic conditions that the target customer group must meet. These conditions collectively constitute a logical "filter," and the core of the system's subsequent work is to estimate how many customers who meet all the conditions of this "filter" will achieve the preset goal (such as purchasing products) in the marketing campaign. Therefore, this step ensures that the prediction task has a clear and unique direction.

[0031] In a preferred embodiment, the system also integrates a light-scale real-time verification and prompting function at this stage. For example, when multiple combinations of conditions selected by business personnel may result in a very small or nonexistent target customer group, the system can provide a real-time prompt message stating that "the combination of conditions may be too strict" based on known customer distribution data. Alternatively, when the system detects that a configured rule is highly similar to a historically successful rule with a high conversion rate, it can proactively display the rule for reference. This real-time interaction not only reduces the error rate of rule configuration but also helps business personnel develop more reasonable marketing strategies more quickly, improving the intelligence and efficiency of human-computer collaboration.

[0032] Step S20: Perform semantic analysis on the business rule information to determine the business association parameters.

[0033] It's important to note that the semantic analysis in this step aims to accurately map the user-configured raw business rules to a series of standard parameters that the system can recognize and process. This process primarily relies on a pre-established business knowledge system. For example, the rules may contain non-standardized business terms such as "VIP customer" or "high-end wealth management." The task of semantic analysis is to match these terms with standard entries in the system's parameter library, such as the "Diamond level" enumeration value under the "Customer Level" parameter, or the "Private Equity Product" enumeration value under the "Product Type" parameter. This ensures that the different colloquial expressions used by different business personnel when inputting rules ultimately point to the same core business concept that can be quantified and calculated.

[0034] Understandably, business-related parameters refer to attribute variables that directly affect the scope and results of customer screening, and whose values ​​can be quantified and represented by the system. Semantic analysis identifies these parameters contained in the original rule text. For example, semantic analysis clarifies the pair of "customer age" parameters, such as "age greater than or equal to thirty years old and less than forty years old," into a numerical range. This numerical range is then used as a constraint condition and bound to the "customer age" parameter, facilitating subsequent standardized processing.

[0035] It should be understood that the semantic analysis process in this step, in addition to basic keyword matching, incorporates logical disambiguation and completeness checks based on the rule's context. For example, when a rule contains both "holding a credit card" and "account assets greater than one million yuan," the system can identify whether there is a logical conflict between these two conditions, classifying them as reasonable parallel filtering criteria. Conversely, if a rule contains an illegal condition such as "account opening date later than the current date," the system will identify and prompt an error at this stage. This process relies on a business knowledge base containing parameter definitions, value ranges, and logical relationship constraints between parameters. This verification process ensures the accuracy and reliability of rule parsing and effectively prevents invalid predictions due to configuration errors.

[0036] Step S30: Classify the business-related parameters according to variable type to obtain the classification results of the business-related parameters.

[0037] It's important to note that the system categorizes identified business-related parameters based on predefined variable type standards. During execution, the system accesses an internally stored parameter type mapping table, which defines the variable type for each system standard parameter. For example, when a business-related parameter is "customer age" or "total account assets," its value can be represented as a continuous numerical value, so the system categorizes it as a "continuous numerical parameter." Parameters like "province" or "product type," whose values ​​come from a finite, discrete predefined list, are categorized as "enumerated parameters." This process is automatic and batch-processed, preparing for the subsequent matching of each parameter to the corresponding feature calculation strategy.

[0038] Understandably, specific variable types can be broadly categorized as follows: Continuous variables, whose values ​​span a continuous range. Examples include "account balance," "customer age," and "transaction amount." This type of data typically contains quantitative information within a certain range. Ordered discrete variables, whose values ​​are discrete, but these values ​​exhibit a clear and comparable order or hierarchical relationship. Although these values ​​themselves may not be directly arithmetically operable, the order is explicit. Examples include "customer risk level" (conservative, moderate, aggressive) or "satisfaction rating" (very dissatisfied, dissatisfied, neutral, satisfied, very satisfied). This type of data retains both categorical and ranking information. Enumerated variables, whose all possible values ​​form a finite and mutually exclusive set, but these values ​​do not have an inherent order, hierarchy, or mathematical relationship; they are merely "labels" used for identification or classification. For example, "gender" (male, female), "product color" (red, blue, green), "city" (Beijing, Shanghai, Guangzhou, Shenzhen), or "occupation" (teacher, engineer, doctor) are characterized by "equality and difference," and any sorting or numerical addition or subtraction operations on them are meaningless. Boolean variables can be seen as a special case of enumeration variables, whose value set has only two elements, usually represented as yes or no. For example, "whether or not you hold a credit card" or "whether you are satisfied with this activity."

[0039] It should be understood that the system can further automatically identify and decompose composite parameters in the aforementioned processing, and determine the type of each component parameter. For example, when the system analyzes the business rule "recently subscribed to funds with an amount greater than 10,000 yuan", the system can identify that the parameter is composed of two more basic parameters, "transaction type" and "transaction amount", which are combined through a logical "AND" relationship. The system will determine "transaction type" as an enumerated parameter and "transaction amount" as a continuous numerical parameter, and process them according to the path corresponding to their respective types. This preprocessing can make parameter classification more refined and accurate.

[0040] Step S40: Calculate the feature value of the business association parameter in the template vector based on the classification result of the business association parameter and the original data in the business rule information.

[0041] It should be noted that the purpose of this step is to perform targeted feature value calculations on the raw numerical or category data extracted from each business rule information, in order to populate the standardized template vector. The system has a built-in feature calculation engine corresponding to each variable type.

[0042] For example, for the parameter "customer age" which has been classified as a continuous numerical type, the engine will directly read the specific age value related to customer age in the original business rule and fill the result as the feature value into the dimension corresponding to "age" in the vector. This process is automated and parallel, ensuring that each business rule can be transformed into a feature vector with a unified structure and machine readability.

[0043] Understandably, the essence of this step is to map diverse and meaningful business parameters into a unified numerical representation through a mathematical model, thereby providing a comparable basis for subsequent rule similarity calculations. The strategy for calculating feature values ​​is closely dependent on the type of parameter. This is because different data types carry information in different ways, and matching mathematical transformations must be used to extract useful information for subsequent analysis without distortion or introducing misleading information.

[0044] Step S50: Construct a prediction template vector based on the feature values ​​of the business association parameters.

[0045] It should be noted that constructing the prediction template vector is the final process of systematically organizing and integrating the multiple feature values ​​obtained from the previous step, which are related to different business parameters. Its aim is to generate a unified, structured numerical representation for each business rule.

[0046] Understandably, the core of this step is to establish a fixed template and fill the designated positions of the template with feature values ​​of different parameters according to preset rules, thereby achieving the "standardization" and "serialization" of the regular features. This ensures that the subsequent prediction model has a unified and unchanging input format. To ensure that the features are aligned with the model input requirements, the originally calculated feature values ​​usually need to be standardized or normalized before filling to eliminate the influence of different units.

[0047] In one embodiment, constructing a prediction template vector based on the feature values ​​of the business-related parameters includes: determining a fixed writing position of the business-related parameters in the prediction template vector according to a preset parameter position index table; determining a fixed number of dimensions of the feature values ​​in the prediction template vector according to the classification type of the business-related parameters; standardizing the feature values ​​of the business-related parameters according to a preset data format; and filling the standardized feature values ​​into a fixed-length vector frame according to the fixed writing position and fixed number of dimensions to obtain the prediction template vector.

[0048] It's important to note that the parameter position index table and the fixed number of dimensions corresponding to the classification type are the basis for this unified transformation process. The former determines which position(s) of each parameter's feature value should be placed in the vector, while the latter determines how many consecutive numerical positions need to be reserved for that parameter. This process ensures the rigor and reproducibility of the construction process. Regardless of the number of rules, as long as the parameter types and definitions remain unchanged, the generated prediction template vectors are completely consistent in length and semantics at each position, laying the foundation for consistent model input. The fixed-length vector frame is the final form of the template, and its length is equal to the sum of the dimensions of all parameter types. For parameter types not mentioned in a certain rule, their corresponding positions are usually filled with zero values ​​or specific placeholders to maintain the integrity of the vector structure.

[0049] Step S60: Input the prediction template vector into the customer number prediction model to obtain the predicted number of qualified customers based on the business rule information.

[0050] It should be noted that the customer number prediction model is a mathematical function built on machine learning technology. Its core function is to automatically estimate the potential number of qualified customers that meet the input business rule characteristics based on the input business rule feature parameters. This model learns the complex non-linear relationship between business rule configuration and the final number of customers by analyzing a large amount of historical marketing campaign data.

[0051] During the model training phase, the system collects various previously executed marketing rules and their corresponding actual number of qualified customers, forming a training sample set. Each historical rule undergoes specific feature engineering, transforming it into a structured numerical vector, which is then paired with its actual customer response count, serving as the foundational material for model learning. The machine learning algorithms employed, such as gradient boosting decision trees, repeatedly traverse these samples, continuously comparing the predicted output with the actual values, gradually adjusting its tens of thousands of internal parameters, ultimately forming a complex function capable of capturing patterns from rule features and making accurate quantity predictions.

[0052] Understandably, the model's input is the feature vector corresponding to the business rule, also known as the prediction template vector. This vector is generated through a standardized feature encoding process. When a business user configures a new rule containing multiple filtering conditions, the system initiates corresponding automated conversion procedures for different types of parameters. For example, for enumeration-type filtering conditions in the rule, such as "account opening city," the system calls pre-calculated and stored business weight data for each city. Based on the specific list of cities selected by this rule, the system extracts the corresponding weight value sequence and immediately calculates multiple key statistical indicators for this sequence. These statistical indicators together constitute a composite feature describing the business value of the "selected city set." Similarly, for numerical conditions, their threshold is directly used as the feature value. Finally, the values ​​obtained after encoding all conditions in the rule are concatenated in a predefined fixed order to form a constant-length, semantically rich numerical array—the feature vector—which serves as the direct input for the model's prediction calculations.

[0053] It should be understood that the construction and application of the entire predictive model follows an end-to-end systematic process. During system deployment and initialization, offline calculations of the business weights for each enumerated parameter are performed based on long-term accumulated historical marketing activity data. Using this as a foundation, batch feature encoding is performed on the historical rule base, thereby constructing a large-scale, high-quality training dataset. Subsequently, appropriate machine learning algorithms are selected to train the model on this dataset. Iterative optimization brings the model parameters to converge, resulting in a stable predictive model. In the online application phase, business personnel configure new rules through an interactive interface. The system backend executes the feature encoding logic in real time, generating corresponding feature vectors and calling the deployed predictive model for forward computation. The model outputs the predicted number of qualified customers within milliseconds, providing immediate and quantitative reference for business decisions. This system forms a closed loop; real-world performance data after new rules are executed is continuously collected and used to periodically update the model and weights, enabling the self-evolution and improvement of predictive capabilities.

[0054] In one embodiment, when the confidence level of the predicted number of qualified customers output by the customer number prediction model is lower than a preset threshold, a degradation processing procedure is initiated.

[0055] It's important to note that when the customer number prediction model calculates a newly configured business rule, in addition to outputting the core numerical result of "predicted number of qualified customers," the system typically generates an internal metric reflecting the reliability of the prediction result: "prediction confidence." This confidence quantifies the model's subjective assessment of its own prediction accuracy. Its calculation may be based on various factors, such as the similarity between the input feature vector and the model's training data distribution (whether it belongs to a rule type the model is familiar with), or the consistency and dispersion of prediction results among the sub-models in the model's ensemble method (such as random forest). To ensure the reliability of the prediction service output, the system pre-sets a confidence threshold. After a prediction calculation is completed, the system automatically checks its confidence. If the value is lower than the preset threshold, the prediction result directly generated by the model is considered high-risk and unreliable. At this point, the system will not return the original model prediction value directly to the business side. Instead, it will automatically trigger a backup and more reliable processing flow, namely the "degradation processing flow". This flow aims to replace the model prediction with a more direct method based on statistical sampling to ensure the availability and a certain degree of accuracy of the final output.

[0056] Understandably, the downgrade process includes: obtaining preset sampling ratio configuration information; extracting sample data from the intermediate data table associated with the business rules according to the sampling ratio configuration information, and constructing a sample data table; matching the sample data table, and taking the number of sample customers that meet the business rules as the sample qualification quantity; determining the estimated qualification quantity of customers based on the sample qualification quantity, the total number of all customers, and the total amount of sample data, and taking the estimated qualification quantity as the predicted qualification quantity of customers.

[0057] The downgrade process uses a classic method of statistical sampling and proportional extrapolation to estimate the number of potential customers. The process first retrieves a preset sampling ratio from the system configuration. This ratio is typically set during system deployment based on historical experience and performance balance, such as 0.1% or 0.01%. Next, the process identifies the target customer group specified by the business rule. The group's identifier is usually stored in an intermediate data table, which has already been pre-filtered based on some non-core screening criteria in the rule. Based on the set sampling ratio, the system randomly or systematically samples from this intermediate data table to extract a statistically representative subset of customers. This subset is used to construct a temporary sample data table. Then, the system applies all screening criteria of the business rule (including all core criteria) to this sample data table, performing a precise data matching query to count the number of sample customers who fully meet all the rules; this is the number of qualified customers. Finally, based on the fundamental principle of simple random sampling, a full projection is performed: the number of qualified customers in the sample is divided by the total sample data to obtain the qualification rate in the sample. This rate is then multiplied by the total number of customers in the intermediate data table to calculate the estimated number of qualified customers. This projection result serves as the final predicted number of qualified customers in the downgrade scenario. Although the entire downgrade process has a higher computational cost than direct model inference, its principles are clear, and the results are easy to interpret. It is an effective safety net strategy when the model cannot provide reliable predictions.

[0058] In this embodiment, by receiving and parsing the business rules configured by the user, classifying the business parameters by type, calculating standardized feature values ​​for each, and filling these heterogeneous feature values ​​into a fixed framework according to preset templates and index rules to form a standard prediction template vector, the vector is finally input into a pre-trained customer number prediction model to obtain the predicted number of customers who meet the rules.

[0059] In summary, this technical solution effectively addresses the problem that complex business rules, due to their diverse parameter types and complex logical combinations, are difficult for machine learning models to process directly. By accurately parsing and standardizing these rules into structurally unified feature vectors, it solves the issue. Because the model can receive highly standardized numerical inputs in both format and semantics, it can more stably and reliably uncover the intrinsic correlation between rule features and customer response volumes, thereby significantly improving the automation level and decision-making accuracy of marketing campaign performance prediction.

[0060] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 In the method for predicting the number of customers meeting the business rules, step S20 includes steps S201 to S204: Step S201: Based on the business rule information, perform basic semantic analysis in conjunction with a predefined business domain dictionary to determine key business terms.

[0061] Understandably, a predefined business domain dictionary refers to a collection of professional terms and common expressions used in various business scenarios within the industry, systematically compiled by relevant business personnel before the prediction behavior begins. Basic semantic analysis refers to the system scanning the received business rule information, using the dictionary for pattern matching to identify and extract key business terms that constitute the core of the rules. For example, from the information "targeting younger demographics among VIP customers," the system can extract the key terms "VIP customers" and "younger demographics" through dictionary matching.

[0062] It should be understood that the purpose of basic semantic analysis is to initially decompose unstructured natural language descriptions or semi-structured configuration information into discrete semantic units that can be further processed. Key business terms refer to the words or phrases that carry core business constraints; they are the concentrated embodiment of the rule's intent. The effectiveness of this step highly depends on the coverage and accuracy of the dictionary. A complete dictionary should encompass all formal names, common aliases, and even industry slang that may be used in the business, ensuring that mainstream expressions are accurately captured.

[0063] Step S202: Based on the key business terms, perform semantic mapping with reference to the parameter definition knowledge base to determine their corresponding standardized parameter names.

[0064] It should be noted that the parameter definition knowledge base stores the standard definitions, unique identifiers, and allowed value ranges or enumeration sets of all processable parameters within the system. Semantic mapping here refers to establishing a correspondence between extracted business terms and standardized parameter names in the knowledge base. For example, the key term "VIP customer" might map to the standardized parameter "customer level," while the term "younger demographic" might map to the standardized parameter "customer age range." This process ensures that different expressions have a unified and unambiguous identifier within the system.

[0065] It is understandable that the same business terminology may refer to different parameters in different contexts. For example, "product" in the context of "holding products" may map to "holding product type," while in the context of "purchasing products" it may map to "target product type." Therefore, in a preferred embodiment, the mapping process is not performed in isolation, but is combined with the overall context of the rule information, and, when necessary, refers to other words before and after the term to assist in decision-making, thereby improving the accuracy of the mapping.

[0066] Step S203: Combine the standardized parameter name with the corresponding value in the business rule information to obtain parameter value pairs.

[0067] In practice, this step integrates and structurally encapsulates the results of the first two steps. After clarifying the standardized parameter name, the system needs to parse the specific numerical value or constraint condition set by the user for the parameter from the original business rule information. A parameter value pair is the smallest semantic unit composed of a standardized parameter name and one or more specific values ​​bound to it. For example, for the parameter "average daily assets of the account" representing a customer's quantitative attribute, its value pair might be "average daily assets of the account: greater than or equal to 10,000 yuan," which is a typical numerical range condition using comparison operators. For the parameter "customer occupation category" representing an enumeration type, its value pair might be "customer occupation category: information technology industry, finance industry," which represents a discrete set of multiple choices. In addition, for time-related parameters, the form of the value pair may be more complex, such as "recent transaction time: within the past thirty days," which reflects a dynamic constraint on the time window.

[0068] The output of this step is a standardized data unit that can be understood and processed by machines. This completes the transformation of the business rules given by the staff from natural language or configuration operations into structured data objects. Each parameter value pair clearly expresses an independent screening condition, which can be used as the most direct input data for subsequent feature vector construction.

[0069] Step S204: Perform logical consistency verification on the parameter value pairs to obtain the business-related parameters.

[0070] In practice, logical consistency verification refers to checking the entire set of parameter value pairs to ensure that there are no contradictory or logically impossible conditions that can be satisfied simultaneously. The verification is based on the constraints between parameters defined in the parameter definition knowledge base. For example, if the system can verify that the value of the parameter "account status" is "closed," it will logically conflict with the requirement that the value of the parameter "recent transaction time" be "within this month," because a closed account should not have any recent transactions.

[0071] In a preferred embodiment, validation includes not only contradiction detection but also integrity alerts. For example, if a rule sets both the "customer risk level" to "conservative" and the "target product type" to "high volatility equity product," the system may not classify it as a hard conflict (as it might theoretically exist), but it can alert the user to the potential low conversion rate risk of this combination based on association rules in the business knowledge base. This proactive, business logic-based validation effectively assists users in formulating more reasonable rules, improving input quality, and thus ensuring the reliability of subsequent predictions. Ultimately, the set of parameter value pairs obtained through all validations is formally determined as the business association parameters used for subsequent feature calculations and predictions.

[0072] In this embodiment, the system identifies the rule content input by the user based on a pre-built business domain dictionary, extracts key business terms, and then uses a parameter definition knowledge base to map each identified term to an internally unique standardized parameter name. These standardized parameter names are then associated and bound with the specific values ​​set by the user in the original rules, thereby forming a series of parameter value pairs. Finally, the system performs an internal logical consistency check on all generated parameter value pairs.

[0073] In summary, the beneficial effects of this embodiment are as follows: First, this solution ensures that similar business rules from different personnel and with different expressions are uniformly and unambiguously parsed within the system, improving the standardization and accuracy of rule processing. Second, the automated and structured processing flow avoids errors and efficiency bottlenecks that may be introduced by manual secondary coding, increasing the speed of rule deployment. Finally, the logical verification mechanism can proactively discover and alert to potential contradictions or inefficient configurations in the rules, enhancing the quality and rationality of the business rules themselves, thereby laying a reliable data foundation for subsequent rule-based high-quality customer feature extraction and behavior prediction.

[0074] Based on the first embodiment of this application, in the third embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 3 In the method for predicting the number of customers meeting the business rules, step S40 includes steps S401 to S405: Step S401: When the classification result of the business association parameter is an enumerated class parameter, obtain the original data of the business association parameter in the business rule information.

[0075] In practice, business rule information is typically submitted by business personnel in a structured manner through a configuration interface. When the system identifies a business-related parameter as an enumeration type, it means that the parameter's value range is a finite set of discrete values.

[0076] It should be noted that the raw data refers to the series of specific enumeration item values ​​selected by business personnel for a certain enumeration type parameter (such as "account opening city") when configuring business rules. In this example, it would be ("City A", "City B", "City C"). The task of this step is to extract the set of specific enumeration items configured for the parameter when the parameter to be processed is identified as an enumeration type.

[0077] It should be understood that the acquisition of raw data must ensure accuracy and consistency. Since business rules may originate from different sources, the descriptions of enumerated items may contain synonyms or differ in format. Therefore, a standardization process is typically required during implementation to ensure accurate matching with the pre-stored enumerated item business weight record table in the system.

[0078] Step S402: Based on the original data of the business association parameters, determine the selected enumeration items included in the original data.

[0079] It's important to note that when configuring rules, business personnel often only specify a subset of enumerated values ​​that are meaningful to their business judgment for a particular parameter. The range of choices they provide is usually incomplete, meaning it doesn't cover all possibilities under that parameter. For example, an anti-fraud rule might only select "high-risk cities," such as cities A and B, as the focus, without listing all risk-free cities. Therefore, "selection" refers to identifying that specific subset, subjectively and intentionally chosen by the rule maker, from the complete set of enumerated values. This step precisely captures and confirms this subset, using it as the sole valid object for subsequent calculations, ensuring that the features only reflect the business intent focused on by the rule maker.

[0080] Step S403: Determine the original weight value of the selected enumeration item based on the enumeration item business weight record table corresponding to the business association parameter.

[0081] Understandably, the system determines the selected enumeration items under a given business parameter by parsing the rule configuration information. This might involve including explicitly checked lists or processing logical groupings or dynamic tags. For instance, a rule might use a tag like "key promotion area," and the system needs to query a predefined tag mapping table to convert it into a list of specifically selected city enumeration items. Each enumeration item thus identified constitutes the list of selected items for that business-related parameter under this rule setting. Its significance lies in excluding other enumeration items not selected in the entire set, which directly affects the sample range considered when calculating the average or other statistics in the next step.

[0082] It should be understood that while different enumeration items themselves have no order, hierarchy, or quantity relationship with other enumeration items, their sample size in the global context does have an overall proportion. For example, for the parameter "account opening city," its possible enumeration items cover all customer account opening locations in the entire system's records. However, the number of customers (or accounts) corresponding to each city enumeration item is different, and their proportions in the total customer base constitute a statistical distribution. Therefore, although "City A" and "City B" are parallel and mutually exclusive discrete options in terms of business scope, their "business weight" or "sample representativeness" may differ significantly due to their different overall proportions. Subsequent steps will require using this "overall proportion" data to weight or calculate the selected enumeration items. For example, a rule might select "Beijing, Shanghai, Guangzhou, and Shenzhen" as target cities. Although they are four equal enumeration items, the final input required by the model might be an aggregated statistic that combines the proportions of these four cities in the global customer base (e.g., 0.15, 0.12, 0.10, and 0.08 respectively), such as a weighted average or the sum of proportions.

[0083] In a preferred embodiment, to ensure global consistency of weight values, this invention establishes and maintains a standardized enumeration item business weight record table from the initial design stage. This record table is an independent static data table that predefines and stores a quantified original weight value for each possible enumeration item under each enumeration parameter in the system. This weight value is usually derived from the statistics of historical business data, such as the proportion of the number of customers corresponding to a specific enumeration item in the total number of customers, or the proportion of the total transaction amount of its related business. When a step requires determining the original weight value of a selected enumeration item, the system does not perform real-time calculation, but directly uses the identifier of the enumeration parameter and the specific code of the selected enumeration item as the precise query key to perform a fast query operation on this weight record table. For example, for the "customer level" parameter, its weight record table may predefine the weight of "ordinary customer" as 0.7, the weight of "VIP customer" as 0.2, and the weight of "strategic customer" as 0.1. The system can directly obtain these values ​​through querying. This design greatly improves processing efficiency and ensures the objectivity of the weights.

[0084] It is important to note that in actual query operations, the completeness of the data in the weight record table must be ensured to match the query key values ​​precisely. If a selected enumeration item has no corresponding record in the weight table, such as adding a new city whose weight has not yet been entered, the system should have a clear fault tolerance and degradation handling mechanism. A common and preferred approach is to first attempt to log this exception and trigger an alert, while assigning a system-predefined default weight value to the missing item, such as the average or median of the weights of all enumeration items under this parameter, to ensure the continuity of the feature calculation process. This avoids process interruption due to missing individual data and provides clues for subsequent data governance through logging.

[0085] Step S404: Calculate the weight feature value of the business association parameter based on the original weight value of the selected enumeration item.

[0086] It should be noted that calculating the weighted feature values ​​of the business-related parameters is a mathematical aggregation process. After the system obtains the original weight values ​​of all selected enumeration items under the business-related accommodation, it will calculate them according to a preset aggregation function. Specifically, it needs to calculate the maximum value, minimum value, average value, standard deviation, and weight sum of these selected enumeration items.

[0087] Understandably, this step aims to integrate the weight information of multiple scattered enumeration items into a single, comprehensive indicator that can holistically represent the business importance of the business parameter under this rule. The magnitude of this calculated weight feature value directly reflects the overall "weight" or "influence" of the set of enumeration items that the rule focuses on within the global business context. For example, a rule for assessing credit rating selects three customer work types—"freelancer," "construction worker," and "individual merchant"—as business-related parameters. If the weight record table shows that the occupational distribution percentages of these three types among the institution's historical non-performing loan customers (i.e., the original weight values) are 0.05, 0.03, and 0.08, then by summing, the maximum, minimum, average, standard deviation, and total weights are 0.08, 0.03, 0.053, 0.16, and 0.025, respectively. This set of weight feature values ​​for the business-related parameter can, to some extent, characterize the multidimensional attributes and confidence level of the rule in the subsequent prediction of the number of qualified customers.

[0088] In one embodiment, calculating the weight feature value of the business association parameter based on the original weight values ​​of the selected enumeration items includes: determining the weight sum and weight extreme values ​​based on the original weight values ​​of each selected enumeration item; calculating the weight average and weight standard deviation formed by the selected enumeration items based on the weight sum; and using the weight sum, weight extreme values, weight average, and weight standard deviation formed by the selected enumeration items included in the business association parameter as the weight feature value of the business association parameter.

[0089] Understandably, this step aims to aggregate the individual weight information of discrete enumerated items into a comprehensive feature vector that can characterize the business characteristics of the rule's focus point from multiple dimensions. The sum of weights represents the total influence or coverage of the selected set of rules within the overall business context. The extreme values ​​of the weights reveal the boundaries and dominant terms of the influencing factors within the rule. The average weight reflects the average business intensity level of the set, while the standard deviation of the weights characterizes the balance of influence among the various factors within the rule.

[0090] Step S405: Obtain the feature value of the business association parameter in the template vector based on the weight feature value.

[0091] In practice, constructing the aforementioned global feature vector is a structured, sequential filling process. First, based on a pre-compiled list of business parameter types, the system assigns a fixed and unique index position to each parameter type. For example, if the system defines twenty different business parameter types, the basic length of the global feature vector is twenty positions. Next, for each business rule to be processed, the system iterates through all parameter types. When a rule involves a specific parameter type, it calls the pre-defined conversion rule for that type (such as the method described earlier for calculating statistics based on enumerated item weights), generating one or more corresponding numerical feature values, and sequentially filling them into the index positions reserved for that parameter type. If a rule does not involve a certain parameter type, a pre-defined default value (such as zero) is filled into that position.

[0092] Understandably, the core advantage of this method lies in achieving homogenized representation of heterogeneous rules through fixed index mapping. Regardless of whether the original rule involves one or more parameters, or how complex its internal logic is, after this process, they are all uniformly transformed into a numerical vector of fixed length and defined semantic positions. The value at each position in the vector represents, in a standardized way, the quantified "contribution" or "strength" of the corresponding parameter type in that rule. This allows thousands of logically distinct rules in complex cases to be organized into a regular two-dimensional matrix, which can then be directly input into subsequent data analysis and machine learning models for processing and learning.

[0093] In a preferred embodiment, to enhance the information carrying capacity of the vector and facilitate tracking, a special field is added at both the beginning and end. Specifically, the starting position of the vector (index zero) is fixed to store the template number on which the rule is based, indicating the feature extraction template or rule paradigm used to generate the vector, facilitating rule classification, tracing, and management. The ending position of the vector (index CNT) is fixed to store the number of qualified customers for the rule. This is a key piece of metadata, reflecting the breadth of coverage or scale of application of the rule in actual business scenarios, providing a direct business volume basis for subsequent model importance weighting or effectiveness evaluation.

[0094] In this embodiment, the method identifies various parameters in the business rules, extracts the selected enumeration items configured by the user from the enumeration type parameters, queries the preset original weight values, and then calculates multiple comprehensive statistics, including the total weight, as its feature values. According to the preset mapping rules, the feature values ​​of all parameter types, together with the rule template number and business usage information, are sequentially filled into a global feature vector with a fixed structure, thereby completing the conversion from business logic to machine-readable data.

[0095] In summary, this embodiment effectively overcomes the technical obstacle of machine learning models directly recognizing and processing different rule features by uniformly transforming business rules with varying meanings into structurally regular numerical vectors. Because feature vectors possess fixed dimensions and semantic positions, the model can efficiently perform batch learning and horizontal comparison of massive amounts of rules in a consistent manner, thereby uncovering deep business patterns hidden beneath complex rule combinations and significantly improving the model training efficiency and knowledge discovery capabilities of intelligent decision-making systems.

[0096] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the method for predicting the number of customers who meet the business rules of this application. Any simple modifications based on this technical concept are within the scope of protection of this application.

[0097] This application also provides a device for predicting the number of customers who meet business rules; please refer to... Figure 4 The device for predicting the number of customers meeting the business rules includes: The business rule formulation module 10 is used to receive business rule information configured by the user. The parameter parsing and classification module 20 is used to perform semantic analysis on the business rule information and determine the business-related parameters. The parameter parsing and classification module 20 is also used to classify the business-related parameters according to variable type to obtain the classification result of the business-related parameters; The feature value calculation module 30 is used to calculate the feature value of the business association parameter in the template vector based on the classification result of the business association parameter and the original data in the business rule information. The model prediction module 40 is used to construct a prediction template vector based on the feature values ​​of the business-related parameters; The model prediction module 40 is also used to input the prediction template vector into the customer number prediction model to obtain the predicted number of qualified customers of the business rule information.

[0098] In one embodiment, the parameter parsing and classification module 20 is further configured to perform basic semantic analysis based on the business rule information and in conjunction with a predefined business domain dictionary to determine key business terms; perform semantic mapping based on the key business terms and with reference to a parameter definition knowledge base to determine their corresponding standardized parameter names; combine the standardized parameter names with the corresponding values ​​in the business rule information to obtain parameter value pairs; and perform logical consistency verification on the parameter value pairs to obtain the business-related parameters.

[0099] In one embodiment, the feature value calculation module 30 is further configured to: obtain the original data of the business association parameter in the business rule information when the classification result of the business association parameter is an enumeration class parameter; determine the selected enumeration item contained in the original data based on the original data of the business association parameter; determine the original weight value of the selected enumeration item based on the enumeration item business weight record table corresponding to the business association parameter; calculate the weight feature value of the business association parameter based on the original weight value of the selected enumeration item; and obtain the feature value of the business association parameter in the template vector based on the weight feature value.

[0100] In one embodiment, the feature value calculation module 30 is further configured to determine the weight sum and weight extreme value based on the original weight values ​​of each selected enumeration item; calculate the weight average value and weight standard deviation formed by the selected enumeration items based on the weight sum; and use the weight sum, weight extreme value, weight average value and weight standard deviation formed by the selected enumeration items included in the business association parameter as the weight feature value of the business association parameter.

[0101] In one embodiment, the model prediction module 40 is further configured to: determine the fixed writing position of the business-related parameter in the prediction template vector according to a preset parameter position index table; determine the fixed number of dimensions of the feature value in the prediction template vector according to the classification type of the business-related parameter; standardize the feature value of the business-related parameter according to a preset data format; and fill the standardized feature value into a fixed-length vector frame according to the fixed writing position and the fixed number of dimensions to obtain the prediction template vector.

[0102] In one embodiment, the model prediction module 40 is further configured to initiate a degradation processing procedure when the prediction confidence of the number of qualified customers output by the customer number prediction model is lower than a preset threshold; the degradation processing procedure includes: obtaining preset sampling ratio configuration information; extracting sample data from an intermediate data table associated with business rules according to the sampling ratio configuration information to construct a sample data table; matching the sample data table to determine the number of sample customers that meet the business rules as the number of qualified samples; determining the estimated number of qualified customers based on the number of qualified samples, the total number of all customers, and the total amount of sample data, and using the estimated number of qualified customers as the number of qualified customers predicted.

[0103] The business rule compliance customer number prediction device provided in this application, employing the business rule compliance customer number prediction method described in the above embodiments, can solve the technical problem of how to quickly and efficiently estimate the number of compliance customers after business rule configuration, while reducing system resource consumption. Compared with the prior art, the beneficial effects of the business rule compliance customer number prediction device provided in this application are the same as those of the business rule compliance customer number prediction method provided in the above embodiments, and other technical features in the business rule compliance customer number prediction device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0104] This application provides a device for predicting the number of customers meeting business rules. The device includes: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores instructions executable by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform the method for predicting the number of customers meeting business rules in Embodiment 1 described above.

[0105] The following is for reference. Figure 5 The diagram illustrates a structural schematic suitable for implementing a business rule compliance customer number prediction device in the embodiments of this application. The business rule compliance customer number prediction device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 5 The business rule-compliant customer number prediction device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0106] like Figure 5 As shown, the business rule compliance customer number prediction device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the business rule compliance customer number prediction device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the business rule compliance customer number prediction device to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows a business rule compliance customer number prediction device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems may be implemented alternatively.

[0107] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0108] The business rule compliance customer number prediction device provided in this application, employing the business rule compliance customer number prediction method described in the above embodiments, can solve the technical problem of how to quickly and efficiently estimate the number of compliance customers after business rule configuration, while reducing system resource consumption. Compared with the prior art, the beneficial effects of the business rule compliance customer number prediction device provided in this application are the same as those of the business rule compliance customer number prediction method provided in the above embodiments, and other technical features of this business rule compliance customer number prediction device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0109] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0110] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0111] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the business rule-compliant customer number prediction method described in the above embodiments.

[0112] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0113] The aforementioned computer-readable storage medium may be included in the business rule compliance customer number prediction device; or it may exist independently and not be assembled into the business rule compliance customer number prediction device.

[0114] The aforementioned computer-readable storage medium carries one or more programs. When these programs are executed by the business rule compliance customer number prediction device, the business rule compliance customer number prediction device performs the following actions: receiving user-configured business rule information; performing semantic analysis on the business rule information to determine business-related parameters; classifying the business-related parameters according to variable type to obtain classification results; calculating the feature values ​​of the business-related parameters in a template vector based on the classification results and the original data in the business rule information; constructing a prediction template vector based on the feature values ​​of the business-related parameters; and inputting the prediction template vector into a customer number prediction model to obtain the predicted number of compliance customers for the business rule information.

[0115] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0116] 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 the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can 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.

[0117] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0118] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described method for predicting the number of customers meeting the business rules. This solves the technical problem of how to quickly and efficiently estimate the number of customers meeting the business rules after configuration, while reducing system resource consumption. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the business rule-compliant customer prediction method provided in the above embodiments, and will not be repeated here.

[0119] The computer program product provided in this application can solve the technical problem of predicting the number of customers who meet business rules. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the business rule-compliant customer number prediction method provided in the above embodiments, and will not be repeated here.

[0120] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for predicting the number of customers meeting business rules, characterized in that, The method for predicting the number of customers meeting the business rules includes: Receive user-configured business rule information; Perform semantic analysis on the business rule information to determine business-related parameters; The business-related parameters are classified according to variable type to obtain the classification results of the business-related parameters; Based on the classification results of the business association parameters and the original data in the business rule information, calculate the feature values ​​of the business association parameters in the template vector; Based on the feature values ​​of the business-related parameters, a prediction template vector is constructed; The prediction template vector is input into the customer number prediction model to obtain the predicted number of qualified customers based on the business rule information.

2. The method for predicting the number of customers meeting business rules according to claim 1, characterized in that, The step of performing semantic analysis on the business rule information to determine business-related parameters includes: Based on the business rule information, and combined with a predefined business domain dictionary, basic semantic analysis is performed to determine key business terms; Based on the key business terms, semantic mapping is performed with reference to the parameter definition knowledge base to determine their corresponding standardized parameter names; The standardized parameter name is combined with the corresponding value in the business rule information to obtain parameter value pairs; Logical consistency verification is performed on the parameter value pairs to obtain the business-related parameters.

3. The method for predicting the number of customers meeting business rules according to claim 1, characterized in that, The step of calculating the feature value of the business association parameter in the template vector based on the classification result of the business association parameter and the original data in the business rule information includes: When the classification result of the business-related parameter is an enumerated parameter, the original data of the business-related parameter in the business rule information is obtained; Based on the original data of the business-related parameters, determine the selected enumeration items contained in the original data; Based on the enumeration item business weight record table corresponding to the business association parameters, determine the original weight value of the selected enumeration item; Calculate the weight feature value of the business association parameter based on the original weight value of the selected enumeration item; Based on the weighted feature values, the feature values ​​of the business association parameters in the template vector are obtained.

4. The method for predicting the number of customers meeting business rules according to claim 3, characterized in that, The step of calculating the weight feature value of the business association parameter based on the original weight value of the selected enumeration item includes: Based on the original weight values ​​of each selected enumeration item, determine the total weight and the extreme value of the weight; Based on the sum of the weights, calculate the average weight and standard deviation of the weights formed by the selected enumeration items; The sum of weights, extreme values ​​of weights, average values ​​of weights, and standard deviation of weights formed by the selected enumeration items included in the business association parameters are used as the weight feature values ​​of the business association parameters.

5. The method for predicting the number of customers meeting business rules according to claim 1, characterized in that, The step of constructing a prediction template vector based on the feature values ​​of the business-related parameters includes: Based on a preset parameter location index table, determine the fixed writing position of the business-related parameter in the prediction template vector; Based on the classification type of the business association parameters, determine the number of fixed dimensions of the feature values ​​in the prediction template vector; The feature values ​​of the business-related parameters are standardized according to a preset data format; The standardized feature values ​​are filled into a fixed-length vector frame according to the fixed writing position and fixed number of dimensions to obtain the prediction template vector.

6. The method for predicting the number of customers meeting business rules according to any one of claims 1 to 5, characterized in that, The method for predicting the number of customers meeting the business rules also includes: When the confidence level of the predicted number of qualified customers output by the customer number prediction model is lower than a preset threshold, the downgrade process is initiated. The downgrade process includes: Obtain the preset sampling ratio configuration information; Based on the sampling ratio configuration information, sample data is extracted from the intermediate data table associated with the business rules to construct a sample data table; The number of sample customers that meet the business rules is used as the number of samples that meet the criteria, based on the sample data table. Based on the number of samples that meet the standards, the total number of customers, and the total amount of sample data, the estimated number of customers that meet the standards is determined, and the estimated number of customers that meet the standards is used as the predicted number of customers that meet the standards.

7. A device for predicting the number of customers meeting business rules, characterized in that, The business rule compliance customer number prediction device includes: The business rule formulation module is used to receive business rule information configured by the user; The parameter parsing and classification module is used to perform semantic analysis on the business rule information and determine the business-related parameters; The parameter parsing and classification module is also used to classify the business-related parameters according to variable type to obtain the classification result of the business-related parameters; The feature value calculation module is used to calculate the feature value of the business association parameter in the template vector based on the classification result of the business association parameter and the original data in the business rule information. The model prediction module is used to construct a prediction template vector based on the feature values ​​of the business-related parameters; The model prediction module is also used to input the prediction template vector into the customer number prediction model to obtain the predicted number of qualified customers of the business rule information.

8. A device for predicting the number of customers meeting business rules, characterized in that, The business rule compliance customer number prediction device includes: a memory, a processor, and a business rule compliance customer number prediction program stored on the memory and executable on the processor, wherein the business rule compliance customer number prediction program is configured to implement the business rule compliance customer number prediction method as described in any one of claims 1 to 6.

9. A storage medium, characterized in that, The storage medium stores a business rule compliance customer number prediction program, which, when executed by a processor, implements the business rule compliance customer number prediction method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the steps of the business rule compliance customer number prediction method as described in any one of claims 1 to 6.

Images