Method and system for generating and operating business credit, device and medium

Abstract

The present disclosure provides a business credit generation and operation method and system, device and medium, the method comprising: S10, collecting the business data required for credit from a multi-source heterogeneous system, and obtaining the basic data for generating credit according to the collected business data. S20, according to the preset credit rules of each level, obtaining the credit data of each level according to the basic data. S30, according to the preset credit audit rules, the credit data of each level is audited respectively, and the audit result is obtained. If the audit result is that the credit data does not pass the audit, trigger the first alarm information and notify the relevant person in charge to review, so that the credit data of each level passes the audit, and the credit data of each level after the audit is obtained, and the credit data of each level after the audit is taken as the basis for calculating the salary of the employee. According to the embodiment of the present disclosure, the credit production standardization process can be constructed, the rule dynamic adaptation and change tracing can be realized, and the problem of rule management confusion can be solved.

Description

Technical Field

[0001] This disclosure relates to the field of business points generation and operation technology, and in particular to a method and apparatus for generating and operating business points, electronic equipment, computer-readable storage medium, and computer program product. Background Technology

[0002] In order to strengthen performance incentives and improve employee enthusiasm, some operators have introduced points-based pay into the compensation structure of front-line employees (points-based pay is where front-line employees add or subtract points based on points rules after performing designated marketing or service tasks, and the accumulated points are converted into compensation according to the compensation rules and sent to front-line employees each month).

[0003] However, in the current field of points production and operation, a complete standardized process system has not been formed. Production mainly relies on the historical production experience accumulated by business personnel. It is impossible to comprehensively audit the data production results and extract effective data production experience from massive orders and complex rules, resulting in the following problems: The production of points systems lacks an integrated, standardized operational process: Enterprises have numerous points categories with complex rules, and there are no standard instructions for these rules. The quality of points rules and data production relies on manual, experience-based analysis, resulting in numerous, time-consuming, and incomplete steps. In traditional production methods, business personnel modify points calculation rules based on periodic rule changes, lacking standardized processes and relying on their understanding of the rules. Rule changes leave no trace, the process is irregular, there are no archives, and subsequent retrieval is difficult.

[0004] The data auditing system for points production is incomplete: Business personnel rely on their understanding of the business and experience with points data to manually diagnose the quality of data production. They can only audit errors with large data fluctuations, and the manual auditing cycle is as long as 2-3 days.

[0005] The generation of visual reports is inefficient: the visual reports for auditing points data rely on manual development and collection by multiple people, and lack intelligent and rapid report generation methods. Summary of the Invention

[0006] This disclosure provides a method and apparatus for generating and operating business points, an electronic device, a computer-readable storage medium, and a computer program product.

[0007] Firstly, this disclosure provides a method for generating and operating business points. This method includes: S10, collecting business data required for points from a multi-source heterogeneous system, and obtaining basic data for generating points based on the collected business data. S20, obtaining point data for each level based on the basic data according to preset point rules for each level. S30, auditing the point data for each level according to preset point audit rules, and obtaining the audit results. If the audit result indicates that the point data fails the audit, a first alarm message is triggered, and the relevant responsible person is notified to conduct a review so that the point data for each level passes the audit, obtaining audited point data for each level, which is then used as the basis for calculating employee compensation. In S10, the business data required for points includes data from tables related to business revenue and expenditure and performance evaluation.

[0008] Optionally, S10 includes: S101, standardizing the format of the collected business data through field regular expression mapping to obtain business data with a unified format; S102, extracting processing time, device MAC address, and service records from the unified format business data, and constructing a user-business association model based on the processing time, device MAC address, and service records; S103, matching offline businesses with online accounts using the user-business association model, completing the association relationship between offline businesses and online accounts, obtaining the completed data, and using the completed data as the basic data for generating points.

[0009] Optionally, S20 further includes: in response to changes in the preset scoring rules for each level, obtaining the changed scoring rules for each level, and using the changed scoring rules for each level to obtain the scoring data for each level based on the basic data. The modifier, modification time, and changes are also recorded.

[0010] Optionally, S30 includes: S301, preprocessing and extracting features from the integral data at each level to obtain extracted features. S302, using an isolated forest model and a K-means clustering model to perform anomaly detection on the extracted features according to preset integral audit rules. The isolated forest model outputs anomaly scores for abnormal integral numbers in the integral data at each level, and the K-means clustering model outputs anomaly scores and business attribution labels for abnormal behavior in the integral data at each level. S303, calling the QWEN large model based on the Transformer architecture, combining it with the integral knowledge base to perform business-oriented reasoning on the business attribution labels, generating reasoning conclusions, and generating an audit report based on the anomaly scores and reasoning conclusions. The audit report includes the problem distribution, anomaly causes, and improvement suggestions. The integral knowledge base includes a preset integral rule library at each level, preset integral audit rules, and a historical anomaly case library. S304, determining the audit results based on the audit report.

[0011] Optionally, the method for generating and operating business points prior to S30 also includes: periodically checking the completeness of basic data and obtaining the check results. If the check results show that the basic data is still not available by the latest deadline, a second alarm message is triggered and the relevant responsible persons are notified to ensure that the basic data is complete.

[0012] Optionally, the methods for generating and operating business points also include: displaying the audited point data at each level in the reporting system according to permissions, generating point reports and pushing them to relevant personnel.

[0013] Optionally, the generation and operation of business points may also include providing customer service to employees through speech recognition, intelligent error correction, intent recognition, knowledge base retrieval, and large-scale models. Customer service includes points inquiry, rule explanation, anomaly feedback, activity consultation, and appeals.

[0014] Optionally, the methods for generating and operating business points also include: constructing an abnormal behavior feature database, and combining a real-time rule engine and audit reports to intercept abnormal behavior in real time during the process of obtaining point data for each level based on basic data, thereby generating risk control logs. Abnormal behaviors include short-term, high-frequency fraudulent orders, point operations by multiple accounts from the same IP address, and abnormal cross-regional business processing.

[0015] Secondly, this disclosure provides a business points generation and operation system, which includes: a business data acquisition module, a points data acquisition module, and an audit module. The business data acquisition module is used to: collect the business data required for points from a multi-source heterogeneous system, and obtain the basic data for generating points based on the collected business data. The business data required for points includes data from tables related to business revenue and expenditure and performance evaluation. The points data acquisition module is used to: obtain the points data for each level according to the preset points rules for each level, based on the basic data. The audit module is used to: audit the points data for each level according to preset points audit rules, and obtain the audit results. If the audit result indicates that the points data fails the audit, a first alarm message is triggered, and the relevant responsible person is notified to conduct a review to ensure that the points data for each level pass the audit, resulting in audited points data for each level, which is then used as the basis for calculating employee compensation.

[0016] Thirdly, this disclosure provides an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores one or more computer programs executable by the at least one processor, the one or more computer programs being executed by the at least one processor to enable the at least one processor to perform the above-described method for generating and operating business points.

[0017] Fourthly, this disclosure provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the above-described method for generating and operating business points.

[0018] Fifthly, this disclosure provides a computer program product that includes computer-readable code or a non-volatile computer-readable storage medium carrying computer-readable code, wherein when the computer-readable code is run in a processor of an electronic device, the processor in the electronic device executes the above-described method for generating and operating business points.

[0019] The embodiments provided in this disclosure can construct a standardized process for points production, realize dynamic rule adaptation and change traceability, and solve the problem of chaotic rule management.

[0020] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0021] The accompanying drawings are provided to further illustrate the present disclosure and form part of the specification. They are used together with the embodiments of the present disclosure to explain the disclosure and do not constitute a limitation thereof. The above and other features and advantages will become more apparent to those skilled in the art from the detailed description of exemplary embodiments with reference to the accompanying drawings, in which: Figure 1 A flowchart illustrating a method for generating and operating business points, provided in this embodiment of the disclosure; Figure 2 A flowchart illustrating another method for generating and operating business points as provided in this disclosure embodiment; Figure 3 A flowchart illustrating yet another method for generating and operating business points as provided in this disclosure embodiment; Figure 4 A schematic diagram of functional modules in an example of a method for generating and operating business points provided in this embodiment of the disclosure; Figure 5 A functional architecture diagram of the points-based question-and-answer function module in an example of a method for generating and operating business points provided in this embodiment of the present disclosure; Figure 6 A structural diagram of the points-based question-and-answer module in an example of a method for generating and operating business points provided in this embodiment of the disclosure; Figure 7 The flowchart illustrates an example of a method for generating and operating business points, provided in this embodiment of the disclosure, including intelligent auditing of points and generation of audit reports. Figure 8A flowchart illustrating a user question-and-answer process in an example of a method for generating and operating business points provided in this embodiment of the disclosure; Figure 9 A block diagram of a business points generation and operation system provided for embodiments of this disclosure; Figure 10 This is a block diagram of an electronic device provided in an embodiment of the present disclosure. Detailed Implementation

[0022] To enable those skilled in the art to better understand the technical solutions of this disclosure, exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments of this disclosure to aid understanding. These should be considered merely exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0023] Where there is no conflict, the various embodiments of this disclosure and the features thereof in the embodiments may be combined with each other.

[0024] As used herein, the term “and / or” includes any and all combinations of one or more related enumerated entries.

[0025] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit this disclosure. As used herein, the singular forms “a” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will also be understood that when the terms “comprising” and / or “made of” are used in this specification, the presence of the stated feature, integral, step, operation, element, and / or component is specified, but the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof is not excluded. Words such as “connected” or “linked” are not limited to physical or mechanical connections but can include electrical connections, whether direct or indirect.

[0026] Unless otherwise specified, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and this disclosure, and will not be interpreted as having an idealized or overly formal meaning, unless expressly so defined herein.

[0027] The method for generating and operating service points according to embodiments of this disclosure can be executed by electronic devices such as terminal devices or servers. Terminal devices can be in-vehicle devices, user equipment (UE), mobile devices, user terminals, terminals, cellular phones, cordless phones, personal digital assistants (PDAs), handheld devices, computing devices, wearable devices, etc. The method can be implemented by a processor calling computer-readable program instructions stored in memory. Alternatively, the method can be executed by a server.

[0028] Figure 1 A flowchart illustrating a method for generating and operating business points according to an embodiment of this disclosure. (Refer to...) Figure 1 The method includes: Step S10: Collect the business data required for integration from the multi-source heterogeneous system, and obtain the basic data for generating integration based on the collected business data.

[0029] In step S10, the business data required for the points system includes data from tables related to business revenue and expenditure and performance evaluation.

[0030] For example, data acquisition technologies such as Secure File Transfer Protocol (SFTP), Shell scripting language, Python programming language, stored procedures, and Structured Query Language (SQL) can be used to collect the required data from relevant databases and store it locally. Data extraction, transformation, and loading (ETL) processes, along with a Business Data Integration (BDI) platform, can be used to integrate the data into a unified acquisition process, and the execution of this process can be monitored. Data in tables related to business revenue and expenditure and performance evaluation includes: tables related to actual revenue increases, pre-deposit amounts, quality penalty deductions, commissioning and testing fees, and initial sales.

[0031] Step S20: According to the preset integration rules for each level, obtain the integration data for each level based on the basic data.

[0032] For example, "each level" refers to the various organizational levels within the company. For instance, point rules can be preset for five levels: company, branch, operations, regional, and individual. These preset point rules might include, for example, double points during the 5G package promotion period. Processing the basic data obtained in step S10, such as calculation, elimination, breakdown, and aggregation, can generate point data for each level. The detailed point data for each level can include business fields such as administrative organization, employee information, business development, development time, point type, point value, and equipment information.

[0033] Step S30: Audit the integral data of each level according to the preset integral audit rules, and obtain the audit results.

[0034] In step S30, if the audit result is that the points data fails the audit, the first alarm message is triggered and the relevant responsible person is notified to conduct a review so that the points data at each level passes the audit and the audited points data at each level is obtained, which is then used as the basis for calculating employee salaries.

[0035] For example, step S30 can be used to analyze integral data, audit integral data problems, and locate the cause of integral anomalies. For example, preset integral audit rules may include: "A daily integral change rate exceeding ±30% is considered abnormal."

[0036] The embodiments of this disclosure provide a method for generating and operating business points. Through steps S10 to S30, the method integrates data from multiple heterogeneous systems, constructs a standardized process for points production, establishes an audit system, and solves the problem of points omission caused by data silos from multiple sources.

[0037] In some embodiments, such as Figure 2 As shown, step S10 includes: Step S101: The collected business data is standardized by using field regular expression mapping to obtain business data with a uniform format.

[0038] For example, field formats of heterogeneous systems such as Business Support System (BSS), Customer Relationship Management (CRM), and Office Automation (OA) can be unified through "field regular expression mapping" (e.g., standardizing the matching of "Business Processing Number" in BSS with "Customer Service Number" in CRM using regular expressions).

[0039] Step S102: Extract processing time, device MAC address, and service record from the standardized business data, and construct a user-business association model based on the processing time, device MAC address, and service record.

[0040] Step S103: Use the user-business association model to match offline businesses with online accounts, complete the association between offline businesses and online accounts, obtain the completed data, and use the completed data as the basic data for generating points.

[0041] For example, a "user-business association model" is constructed based on step S101. Then, based on the processing time, device MAC address, service records and other dimensions obtained in step S102, the association between offline business and online account is automatically completed (e.g., when a 5G package is processed offline but not associated with an online account, the user's online ID is matched through device information), reducing the points under-counting rate from 20% to below 2%.

[0042] In some embodiments, step S20 further includes: in response to a change in the preset scoring rules for each level, obtaining the changed scoring rules for each level, and using the changed scoring rules for each level to obtain the scoring data for each level based on the basic data. The modifier, modification time, and changes are also recorded.

[0043] For example, the method disclosed herein supports hot updates of points rules. After a rule change, the modifier, the time of modification, and the content of the change are automatically recorded, enabling full-process traceability.

[0044] In some embodiments, such as Figure 3 As shown, step S30 includes: Step S301: Preprocess and extract features from the integral data at each level to obtain the extracted features.

[0045] For example, preprocessing mainly includes cleaning (e.g., removing null values, format conversion) and standardization (e.g., code mapping dictionary). Extractable features include four categories: user dimensions (user ID, organization), time series (daily points month-on-month / year-on-year change, amount of change), behavioral features (frequency of points acquisition, number of accounts associated with a device), and business features (types of bonus points, types of deduction points).

[0046] Step S302: Based on the preset integral audit rules, use the isolated forest model and the Kmeans clustering model from step S to perform anomaly detection on the extracted features.

[0047] In step S302, the isolated forest model outputs the anomaly scores for abnormal integral numbers in the integral data of each level, and the Kmeans clustering model in step S outputs the anomaly scores and business attribution labels for abnormal behavior in the integral data of each level.

[0048] Understandably, step S302, based on intelligent analysis of integral data from small models, can quickly identify integral features and perform anomaly detection, preparing sufficient corpus information for large models to improve their inference accuracy. Small models include isolated forests, K-means clustering, etc.

[0049] For example, based on preset points audit rules, an isolated forest model can be used to detect abnormal point values ​​(such as an employee's points for the billing period suddenly increasing from 3 to 10, a change rate of 700%), and a K-means clustering model can be used to identify abnormal behavior (such as the same device being associated with 5 employee accounts to conduct business within 30 minutes). The model can then output "abnormal score + business attribution label" (such as "multiple accounts associated with the device - suspected points fraud" or "offline business not associated with online - suspected omission"), providing accurate data for the large model.

[0050] Taking an abnormal surge in points as an example, a user's recent account balance is collected as {1,2,1,3,10}, with a sudden increase at point 10. During the points information collection process, the points sequence {1,2,1,3,10} is collected. During the points feature extraction process, the recent account balance change rate index is extracted as {1,-1,2,7}, where 7 represents an abnormal surge. In the intelligent analysis phase of the points data, the average daily points change rate is set to 0, the standard deviation to 1, and the range three times the standard deviation to {-3,3}. Therefore, when the account balance change of 7 exceeds the threshold range, it is defined as an outlier.

[0051] Step S303: Call the QWEN large model based on the Transformer architecture, combine the integral knowledge base to perform business-oriented reasoning on the business attribution tags, generate reasoning conclusions, and generate an audit report based on the anomaly score and reasoning conclusions. The audit report includes the problem distribution, anomaly causes and improvement suggestions.

[0052] In step S303, the points knowledge base includes a pre-set points rule base for each level, pre-set points audit rules, and a historical anomaly case base.

[0053] For example, the large model is fine-tuned using the QWEN framework based on the Transformer architecture. The points knowledge base is called to perform business-oriented reasoning on the abnormal labels output by the small model. For example, for the label "multiple accounts associated with a device", the reasoning conclusion is "the device is associated with 5 accounts to conduct the same business on the same day, which is consistent with the characteristics of 'points farming' behavior. It is recommended to intercept the corresponding points and verify the account ownership".

[0054] For example, the reasoning conclusions (problem distribution, anomaly causes, and improvement suggestions) can be generated into Word / PDF / Excel format reports using technologies such as python-docx (Python for manipulating Word documents), Apache POI (Java for manipulating Office documents), and xlsxwriter (Python for manipulating Excel files). These reports can be viewed and downloaded online by relevant personnel. The reports include business conclusions (such as "3 abnormal points in XX package area originated from offline business not being associated with online accounts, and the mapping logic between the OA and BSS systems needs to be optimized").

[0055] Step S304: Determine the audit results based on the audit report.

[0056] Understandably, the audit result is either the integral data failed the audit or the integral data passed the audit.

[0057] In some embodiments, before step S30, the method further includes: periodically checking the completeness of the basic data and obtaining the detection results. If the detection results show that the basic data is still not available when the latest time limit is reached, a second alarm message is triggered and the relevant responsible person is notified to ensure that the basic data is complete.

[0058] Understandably, basic data can be generated periodically, such as according to the payroll calculation cycle (or payment period). Typically, the payroll calculation cycle is one month, and basic data can be generated monthly.

[0059] For example, basic data detection rules can be pre-configured, such as "the actual data collected for the previous month must be completed before 9:00 AM on the 1st of each month" and "the integrity verification of basic data (such as no missing actual data or pre-stored data) must be completed before points production." The implementation can be as follows: a Java timer class (such as Timer) combined with the distributed task scheduling component XJob timer task component is used to periodically perform basic data detection (e.g., every 30 minutes). If the basic data is still not available by the latest deadline (e.g., 12:00 PM on the 1st of each month), an interface alarm is triggered (displaying "XX billing period basic data not collected, please check the upstream system" on the system alarm page) and an SMS notification is sent (calling the SMS gateway to open the Application Programming Interface (API) interface to send a second alarm message to the responsible person's mobile phone, such as "[XX Points System] XX billing period basic data not collected, responsible person: Zhang San, contact information: 138XXXX1234"); the detection is repeated until the basic data is available.

[0060] In some embodiments, the method for generating and operating business points further includes: displaying the audited point data at each level in the reporting system according to the hierarchical access permissions, generating point reports and pushing them to relevant personnel.

[0061] Understandably, a hierarchical display method based on permissions ensures that users at different levels can only view data within their authorized scope. For example, a reporting system can be built using Java programming language, Spring Boot framework, HyperText Markup Language 5 (HTML5), and Vue.js 3 front-end framework. This system supports five levels of data display: branch office, service center, zone, and individual. Permissions are assigned based on user roles (e.g., zone managers can only view their own and their zone's data, while branch office administrators can view all branch office data). Reports can include details of points, point summaries, and statistics on abnormal points. Data can be filtered (by time and business type) and exported.

[0062] For example, points data can be pushed to employees through enterprise communication applications (such as WeChat Work) to achieve real-time delivery of points information. For instance, a points report HyperText Markup Language (HTML) page (containing points for the current period, year-on-year and month-on-month changes, and anomaly alerts) can be generated, and the WeChat Work API interface can be called to push it to the user's WeChat Work account according to user permissions. Users can log in to WeChat Work and click the link to view it. Timed pushes (such as pushing the previous month's points report on the 5th of each month) and anomaly pushes (such as pushing alarm information in real time when points audit fails) can be implemented.

[0063] In some embodiments, the method for generating and operating business points further includes providing customer service to employees through speech recognition, intelligent error correction, intent recognition, knowledge base retrieval, and large-scale models. Customer service includes points inquiry, rule explanation, anomaly feedback, activity inquiries, and appeals.

[0064] For example, the human-computer interaction process when providing customer service is as follows: User Questions: Receives voice or text input from users.

[0065] Speech recognition: Converting speech signals into readable text. Speech recognition can use deep learning speech recognition models, such as the wav2vec 2.0 speech recognition model, which can convert user speech questions into text, remove silence, reduce noise, and unify punctuation and capitalization.

[0066] Intelligent error correction: Perform semantic error correction and spelling correction on the recognized text, and output standardized error-free text. For example, a BERT language model (i.e., a bidirectional encoder based on Transformer) + a custom integral term error correction dictionary (such as correcting "integral points" to "integral", "5G package parameters" to "5G package") can be used to correct typos and pronunciation-confused words.

[0067] Intent recognition: Recognize the user's intent, return if it does not conform to the integral-related intent, and extract key information according to the intent for the next step of reasoning; the knowledge base provides knowledge support for intent understanding and result generation. For example, by fine-tuning the BERT model, 5 types of integral-exclusive intents can be recognized: integral query type ("How many points do I have this month"), rule explanation type ("Why is the 5G package double points"), exception feedback type ("Why did my points decrease by 20"), activity consultation type ("How many points does a new user get for applying for a card"), appeal type ("Need to appeal for incorrect integral calculation"), and multi-round conversation memory is supported (such as when the user first asks "5G package integral rules" and then asks "Why don't I get double points for applying for 5G", the system automatically associates the previous conversation).

[0068] Structural reasoning: Generate a structured answer by combining the user's intent, key information, user questions, knowledge base, etc. For example, by combining the user's intent and historical behavior (such as the user's package area and historical service handling records), recall the top 3 (Top-3) relevant knowledge (such as the 5G package integral rules of the user's package area) from the integral knowledge base through a vectorized search engine (such as the Milvus vector database), and sort by relevance + personalization.

[0069] Result output: Present the reasoning result to the business personnel in the form of text or voice. For example, the fine-tuned integral large model combines the user's question, intent label, and retrieved knowledge to generate a structured natural language answer. For example, when the user asks "Why did my points decrease by 20", the answer includes "You had 20 points deducted this month due to 'Service quality deduction' (corresponding service: Customer complaint handling timeout on March 15, 2024). Deduction rule: 20 points are deducted for each service complaint, and points can be recovered by applying for 'High-quality service supplementary recording'."

[0070] In some embodiments, the method for generating and operating business points further includes: constructing an abnormal behavior feature library, and in the process of obtaining the integral data of each level according to the basic data, combining a real-time rule engine and an audit report to intercept abnormal behaviors in real time and generate a risk control log. Abnormal behaviors include behaviors of short-term high-frequency false orders, behaviors of integral operations of multiple accounts with the same IP, and abnormal behaviors of cross-regional service handling.

[0071] For example, the method for constructing the abnormal behavior feature database can be as follows: More than 10 types of abnormal points-based behaviors are identified and a feature database is built, including "short-term, high-frequency fraudulent orders (e.g., processing more than 10 identical transactions within 1 hour)," "multiple accounts operating points from the same IP address (e.g., logging into more than 5 employee accounts from the same IP address in a single day to operate points)," and "abnormal cross-regional business processing (e.g., an employee handling business in area B from area A)." The real-time interception method can be as follows: combining a real-time rule engine with audit reports, when an abnormal behavior feature database rule is triggered during points generation (e.g., logging into 6 employee accounts from the same IP address in a single day to operate points), the corresponding points are automatically intercepted (e.g., no points are generated), and a risk control log is generated (recording the interception time, involved accounts, abnormal behavior type, and intercepted point value). With an interception rate ≥ 95%, annual points losses can be reduced by over 5 million yuan.

[0072] In some embodiments, the method for generating and operating business points further includes: monitoring various operational metrics and corresponding thresholds, and triggering an alarm and notifying relevant personnel when the operational metrics exceed the corresponding thresholds.

[0073] For example, the operational metrics and corresponding thresholds include the ETL data acquisition task success rate (with a threshold of 99%), the response time of the intelligent audit module (with a threshold of 5 seconds), and the large model inference accuracy (with a threshold of 95%). When a monitored metric exceeds a threshold (e.g., the ETL task success rate drops to 95%), an alarm is triggered, and alarm information (e.g., "ETL acquisition task success rate is abnormal, current value 95%, threshold ≥ 99%) is displayed on the system monitoring interface, and technical maintenance personnel are notified (e.g., alarm information is pushed via WeChat).

[0074] The business points generation and operation method disclosed in this embodiment constructs a standardized process for points production, realizes dynamic rule adaptation and change traceability, and meets the requirement of second-level effectiveness for more than 30% of rule changes per quarter, solving the problem of chaotic rule management. It establishes an intelligent audit system of "small model anomaly detection + large model business reasoning" to discover the relationship between rules, orders, and points, shortening the audit cycle from 2-3 days to hours and improving the accuracy rate to over 98%. Simultaneously, it generates a visual report containing business conclusions and improvement suggestions to support operational decision-making. It achieves data fusion from multiple heterogeneous systems to solve the problem of points under-calculation, establishes a real-time risk control mechanism (interception rate ≥95%) to intercept points fraud, and builds an intelligent points Q&A system to accurately identify employee intentions and provide real-time answers, reducing the workload of manual customer service by more than 80%.

[0075] The following example illustrates the method for generating and operating business points provided in this disclosure.

[0076] like Figure 4As shown, this example includes 6 core functional modules. These modules will be introduced below: 1. Points Collection and Production Module: This module is used to generate the basic data and monthly points data upon which the points production of this invention depends, serving as the basis for calculating the points-based compensation of frontline employees each month and solving the problem of points omission caused by data silos from multiple sources.

[0077] Specific implementation: (1) Data collection: Select the business data required for points from the public model issued by China Unicom Group and the Beijing Unicom data warehouse (such as the table related to actual revenue increase, the table related to pre-deposit, the table related to quality deduction, the table related to commissioning fee, and the table related to first sale). Use SFTP / Shell / Python scripts, stored procedures, SQL and other technologies to collect data and store it locally. Use ETL and BDI (data governance platform) to integrate the program to form a unified collection process and monitor the execution of the process.

[0078] (2) Multi-source data fusion: Unify the field formats of heterogeneous systems such as BSS, CRM, and OA through "field regular expression mapping" (e.g., standardize the matching of BSS "business processing number" and CRM "customer service number" by regular expression); construct a "user-business association model" to automatically complete the association between offline business and online account based on dimensions such as processing time, device MAC address, and service records (e.g., when a 5G package is processed offline but not associated with an online account, the user's online ID is matched through device information), reducing the points omission rate from 20% to below 2%.

[0079] (3) Points-based production: According to the five-level points rules of company, branch, service, area, and individual, the basic data after cleaning and merging is processed by calculation, elimination, disassembly, and aggregation to generate points data at each level. It supports hot rule updates and automatically records the modifier, modification time, and change content after rule changes, so as to achieve full-process traceability.

[0080] Technical implementation: SQL, Shell, and Python programming; ETL and BDI data governance platform; field regular expression mapping algorithm; user-business association matching model.

[0081] 2. Intelligent Audit Module for Points: This module is mainly used to intelligently analyze points data, audit points data problems, and locate the reasons for large changes in points. It focuses on anomaly detection and business-oriented reasoning of points business characteristics.

[0082] Specific implementation: (1) Collection of points information: Use ETL tools to collect points production data, configure points definition rules (such as "double points during the 5G package promotion period") and points audit rules (such as "a daily points change rate exceeding ±30% is abnormal"). The points data details include business fields such as administrative organization, employee information, business development, development time, points type, points value, and equipment information.

[0083] (2) Extraction of points features: The collected data is cleaned (removing null values ​​and converting the format) and standardized (code dictionary mapping) to extract four categories of features: user dimension (user ID, organization), time series (daily points month-on-month / year-on-year, change), behavioral features (frequency of points acquisition, number of accounts associated with the device), and business features (type of business that adds points, type of business that deducts points).

[0084] (3) Small model intelligent analysis: The isolated forest model is used to detect abnormal score values ​​(such as an employee's credit period score suddenly increasing from 3 to 10, with a change rate of 700%), and the Kmeans clustering model is used to identify abnormal behavior (such as the same device being associated with 5 employee accounts to conduct business within 30 minutes). The output is "abnormal score + business attribution label" (such as "multiple accounts associated with the device - suspected score fraud" and "offline business not associated with online - suspected omission"), providing accurate data for the large model.

[0085] (4) Large model reasoning audit: QWEN based on the Transformer architecture is used to fine-tune the large model, and the points knowledge base (including the rule base and historical abnormal case base) is called to perform business-oriented reasoning on the abnormal tags output by the small model. For example, for the tag "device multiple account association", the reasoning conclusion is "the device is associated with 5 accounts to handle the same business on the same day, which is consistent with the characteristics of 'points farming' behavior. It is recommended to intercept the corresponding points and check the account ownership".

[0086] (5) Audit report generation: Using python-docx, Apache POI, and xlsxwriter technologies, the audit results (problem distribution, cause of anomalies, and improvement suggestions) are generated into Word / PDF / Excel format reports, which can be viewed and downloaded online. The reports include business conclusions (such as "3 abnormal points in XX package area are due to offline business not being associated with online accounts, and the mapping logic between OA and BSS systems needs to be optimized").

[0087] Technical implementation: Isolation Forest, Kmeans clustering small model, QWEN fine-tuning large model, python-docx, Apache POI, xlsxwriter document processing technology, integral knowledge base (graph database storage rules, vector database storage case).

[0088] 3. Report Display Module: Mainly used to display the newly generated integral data, implement refined control of data permissions, and ensure that users at different levels can only view data within their permission scope.

[0089] Specific implementation: Build a report system through Java, Spring Boot, HTML5, and VUE3 technologies, support the display of 5-level data of branch companies, business operation services, package areas, and individuals. Allocate permissions based on user roles (for example, the package area head can only view the data of himself and the affiliated package area, and the branch company administrator can view the data of the entire branch company). The report includes modules such as integral details, integral summary, and abnormal integral statistics, and supports data screening (by time, business type) and export functions.

[0090] 4. Report Express Module: Used to push integral data through enterprise WeChat to achieve real-time reach of integral information. Specific implementation: Generate an integral express HTML page (including integral for this accounting period, year-on-year and month-on-month changes, and abnormal prompts), call the enterprise WeChat open API interface, and push it to the corresponding enterprise WeChat account according to user permissions. Users can view it by logging in to enterprise WeChat and clicking on the link, and support scheduled push (such as pushing the integral express of the previous month on the 5th of each month) and abnormal push (such as real-time pushing of an alarm express when the integral audit fails).

[0091] Technical implementation: HTML5 page development, enterprise WeChat API interface docking, scheduled task scheduling (XJob component).

[0092] 5. Integral Q&A Module: Used to solve the integral problems of front-line personnel, design a multi-modal memory mechanism, and improve the accuracy and coherence of Q&A.

[0093] The specific implementation is as Figure 5 shown, and it includes the following parts: (1) Data preprocessing: Clean the integral data (such as standardizing the integral type name), and preprocess the Q&A data (unify the question wording format, for example, "integral deduction" and "reason for deduction item" are uniformly classified as abnormal feedback categories).

[0094] (2) Speech recognition: Adopt the wav2vec2 deep learning model to convert the user's speech question into text, remove silence and noise, and unify punctuation marks and capitalization.

[0095] (3) Intelligent error correction: Based on the BERT language model + custom integral term error correction dictionary (such as correcting "integral points" to "integral", and "5G set parameters" to "5G package"), correct typos and pronunciation confusion words.

[0096] (4) Intent recognition: By fine-tuning the BERT model, five types of points-specific intents are identified: points query ("How many points do I have this month?"), rule explanation ("Why is the 5G package double points?"), anomaly feedback ("Why are my points 20 points less?"), activity consultation ("How many points do new users get when they sign up for a card?"), and appeal ("Points calculation error, appeal needed"). Multi-turn dialogue memory is supported (e.g., if a user first asks "5G package points rules" and then asks "Why am I not getting double points after signing up for 5G?", the system will automatically link the previous dialogue).

[0097] (5) Points knowledge base for knowledge retrieval: Combining user intent and historical behavior (such as the user's package area and historical business processing records), the top-3 relevant knowledge (such as the 5G package points details of the user's package area) is retrieved from the points knowledge base through a vectorized search engine (such as Milvus), and sorted by relevance + personalization.

[0098] (6) The large-scale points model is used to generate answers: The fine-tuned large-scale points model combines user questions, intent tags, and retrieval knowledge to generate structured natural language answers. For example, if a user asks "Why did I lose 20 points?", the answer includes "You were deducted 20 points this month due to 'service quality penalty' (corresponding business: customer complaint processing timeout on March 15, 2024). Penalty rule: 20 points are deducted for each service complaint. Points can be recovered through 'high-quality service supplement'."

[0099] Technical implementation: wav2vec2 speech recognition model, BERT fine-tuned intent recognition model, Milvus vectorized search engine, QWEN integral domain large model, multi-turn dialogue context memory algorithm.

[0100] 6. Monitoring and Alarm Module: Monitors the system's operating status and abnormal points-based behavior, enabling production condition detection and real-time risk control interception, and adds dedicated risk control capabilities for points-based business.

[0101] Specific implementation: (1) Production monitoring: - Detection rule configuration: Pre-configure basic data detection rules in the system, such as "the actual data received in the previous month must be collected before 9:00 on the 1st of each month" and "the basic data integrity verification (such as no missing actual data or pre-stored data) must be completed before points production"; - Scheduled Detection: Utilizing Java timer classes (such as Timer) combined with the XJob scheduled task component, basic data detection is performed periodically (e.g., every 30 minutes). If the basic data is still unavailable by the latest deadline (e.g., 12:00 PM on the 1st of each month), an interface alarm is triggered (displaying "Basic data for XX billing period not collected, please check upstream system" on the system alarm page) and an SMS notification is sent (calling the SMS gateway API to send the alarm information to the responsible person's mobile phone, such as "[Beijing Unicom Points System] Basic data for XX billing period not collected, responsible person: Zhang San, contact number: 138XXXX1234"). The detection is repeated until the basic data is available. (2) Risk control interception: - Construction of an abnormal behavior feature library: Identify 10+ types of abnormal points-based behaviors and build a feature library, including "short-term high-frequency fake orders (e.g., processing more than 10 identical transactions within 1 hour)", "multiple accounts operating points from the same IP (e.g., logging into more than 5 employee accounts from the same IP address to operate points in a single day)", and "abnormal cross-regional business processing (e.g., an employee processing business in area B from area A)". - Real-time interception: Combining the anomaly detection results of the real-time rule engine and the intelligent points audit module, when an abnormal behavior feature library rule is triggered during the points production process (such as the same IP address logging into 6 employee accounts in a single day to operate points), the system automatically intercepts the corresponding points (if the points are not generated), and generates a risk control log (recording the interception time, involved accounts, abnormal behavior type, and intercepted points value); the interception rate is ≥95%, reducing points losses by more than 5 million yuan per year; (3) System monitoring: - Monitoring indicator configuration: Configure the monitoring indicators for each module of the system, such as the success rate of ETL data acquisition tasks (threshold ≥ 99%), the response time of the intelligent audit module for points (threshold ≤ 5 seconds), and the accuracy of large model inference (threshold ≥ 95%). - Alarm Trigger: When the monitored metric exceeds the threshold (e.g., ETL task success rate drops to 95%), an alarm is triggered, and alarm information is displayed on the system monitoring interface (e.g., "ETL collection task success rate is abnormal, current value is 95%, threshold ≥ 99%)", and technical maintenance personnel are notified (e.g., alarm information is pushed via WeChat).

[0102] Technical support: Java timer technology, XJob component, SMS gateway API, real-time rule engine, abnormal behavior feature library, risk control interception algorithm.

[0103] like Figure 6 As shown in the diagram, the flowchart for points generation and operation in this example mainly includes 7 steps.

[0104] Step S1: Basic Data Collection for Points System. Using technologies such as SFTP, shell scripting, Python, stored procedures, and SQL, the necessary business data for points is extracted from the public models issued by China Unicom Group and the Beijing Unicom data warehouse. This includes tables related to actual revenue increases, pre-deposits, quality penalties, commissioning fees, and first-sales transactions. This data is then stored locally. After data cleaning, processing, and consolidation, the basic data for points production is formed. ETL and BDI (Data Governance Platform) are used to integrate these programs or scripts into a unified process, controlling the execution of the entire process and monitoring and auditing its execution.

[0105] Step S2: Check if basic data is available. The monitoring and alarm module will automatically determine whether the basic data for the current billing period is available based on the pre-configured basic data detection rules in the system. If the basic data is still not available by the latest acceptable time, an on-screen alarm and SMS notification will be issued.

[0106] The monitoring and alarm module will periodically execute step S1, as described above, and repeat this cycle until the underlying data is unknown.

[0107] Interface-based alarms display alarm information on the system's alarm page, which frontline personnel can see upon logging into the system. SMS notifications send alarm information to each person's mobile phone, reminding relevant personnel to investigate the reason why the upstream system failed to provide data on time.

[0108] This part requires the use of timing technology. Java's standard classes have relevant timing methods, and the xjob component can be used for better implementation of scheduled tasks. UI alerts can be implemented using traditional front-end and back-end development technologies, which have been listed above and will not be elaborated here. SMS functionality involves communication with SMS gateways or SMS API interfaces.

[0109] Step S3: Points Data Generation. Based on the points rules for companies, branches, service centers, service areas, and individuals, the basic data undergoes processing operations such as calculation, elimination, breakdown, and aggregation to generate points data for each level. The technology used is similar to that in Step S1, only the algorithm differs.

[0110] Step S4: Intelligent auditing of integrators and generation of audit reports. This part utilizes the data analysis and mining capabilities of a small model and the logical reasoning capabilities of a large model, such as... Figure 7 As shown, the specific algorithm is as follows: Step S4.1: Points Information Acquisition Module. Use an ETL tool to collect the raw points data generated in step S3 and perform further data processing; configure points definition rules, points verification rules, and other points information in the system.

[0111] The main attributes of the points data details include administrative organization, employee information, business development, development time, points type, and points value. Points definition rules refer to points policies, points rule definitions, and other points rules based on business experience.

[0112] Integral audit rules refer to the rules for monitoring the rationality of the generated integral data, including rules on the magnitude of integral changes and the rationality of integral changes.

[0113] Step S4.2: Integral Feature Extraction. Extract the collected integral information and perform preprocessing operations such as data cleaning and standardization, as well as feature engineering operations such as feature extraction.

[0114] Data cleaning methods include removing null values ​​and converting data formats. Data standardization methods include code dictionary mapping, etc. Feature extraction includes user-dimensional feature extraction (e.g., user ID, organization, etc.), time series feature extraction (e.g., daily points month-on-month, year-on-year, change, points time, etc.), behavioral feature extraction (e.g., frequency of points acquisition, average points, etc.), and business feature extraction (e.g., business that increases points, business that decreases points, etc.).

[0115] Step S4.3: Intelligent analysis of integral data based on small models. This is used to quickly discover integral features and perform anomaly detection, preparing sufficient corpus information for the large model to improve its inference accuracy. Small models include Isolation Forest, K-means clustering, etc. In terms of feature engineering, integral data is input and integral features are output; in terms of anomaly detection, feature vectors are input and anomaly scores are output, and anomaly samples are filtered based on the set predictions.

[0116] Below is an example of a small model identifying abnormal points. Taking a sudden increase in points as an example, a user's recent account balance is collected as {1,2,1,3,10}, with a sudden increase at point 10. During the points information collection process, the points sequence {1,2,1,3,10} is collected. During the points feature extraction process, the recent account balance change rate index is extracted as {1, -1,2,7}, where 7 represents an abnormal increase. In the intelligent analysis phase of the points data, the average daily points change rate is set to 0, the standard deviation to 1, and the range three times the standard deviation to {-3,3}. Therefore, when the account balance change of 7 exceeds the threshold range, it is defined as an abnormal value.

[0117] Step S4.4: Intelligent review of integration results based on a large model. This step is used to generate an integration review report. The large model used is the integration large model, which performs model inference and integration on the output results of the small model to output the review result report.

[0118] The large model used here is the same as the large model used in step S7, and will not be repeated here.

[0119] Step S4.5: Generate an audit report. The audit report includes the distribution of issues, audit results and recommendations, etc. The report format can be PDF, Excel, XML, etc., and a download function is provided.

[0120] Document manipulation requires the use of document manipulation technologies such as python-docx or Apache POI. Download functionality can be implemented using traditional front-end and back-end technologies; the key is to implement robust system security controls and data access control to prevent data leaks.

[0121] Step S5: Determine if the audit passes. This part determines whether the audit result passes. If it fails, an alarm message needs to be generated, displayed on the interface, and sent to relevant personnel via SMS. This part is similar to the function of step S2, both being monitoring and alarm module functions, but the algorithms and business processing logic used are different.

[0122] Step S6: Display of points-related reports and push notifications for points updates. After step S5 determines that the audit has passed, a points report will be generated and displayed in the report display module. At the same time, a points update will be generated and sent to the relevant personnel's WeChat account.

[0123] The report display module can be implemented using traditional front-end and back-end development technologies. For Enterprise WeChat push, it is necessary to be familiar with the API interface specifications of Enterprise WeChat. After debugging with back-end technology, the Enterprise WeChat push of points report can be implemented.

[0124] Step S7: Intelligent Answering of Points-Based Questions. When frontline staff see their points for the current billing period, especially when there are deductions, they need to know how the points were deducted. Points-based questions can then provide detailed explanations of the reasons for these deductions. Simultaneously, it can clarify the composition of frontline staff's points for the current billing period and inform them of the points rules—what behaviors earn what points—thus motivating frontline staff.

[0125] The implementation principle of the points-based question-and-answer system is as follows: 1. Data Preprocessing. This module preprocesses the collected data, such as cleaning and standardizing the collected points data; and preprocessing the collected question-and-answer data to meet contextual requirements.

[0126] 2. Speech Recognition Module. This module processes user questions in speech form, using a speech-to-text model to convert the original speech file into text. Its main functions include: speech processing (removing silence); noise reduction; formatting (converting to a common format); and text processing (standardizing punctuation, capitalization, etc.).

[0127] 3. Intelligent Error Correction Module. This module corrects user-submitted questions by detecting and correcting typos, mispronunciations, and misspellings, and by providing contextual corrections. It utilizes language models such as BERT, a custom error correction dictionary, and a rule engine for error correction.

[0128] 4. Intent Recognition Module. This module determines the category of user intent, identifying the user's question intent based on lightweight models such as BERT. Intents include points queries (How many points do I have?), rule explanations (Why did I get 10 points?), error feedback (Why are my points 2 instead of 10?), activity inquiries (How many points should I get for this activity?), and appeals (My points are wrong). It supports multi-turn dialogues and contextual understanding, improving the accuracy of points-based question-and-answer sessions.

[0129] 5. Income Points Knowledge Base. This module's income points knowledge base draws from income points rules (acquisition conditions, etc.), income points question-and-answer data, and income points policy explanations (frequently asked questions). Relevant data is preprocessed and integrated to form an income points question-and-answer knowledge base, providing contextual support for intent recognition and a basis for knowledge retrieval for large models. The knowledge base is stored in graph databases, vector databases, or document databases, depending on the data source.

[0130] 6. Integral Large-Scale Model Module. This module employs a pre-trained large-scale model based on the transformer architecture. It utilizes an integral knowledge base for fine-tuning, optimizing model parameters and improving the model's ability to understand integral information. The fine-tuned integral domain large-scale model is deployed, leveraging its deep feature extraction capabilities to generate the final answer. Inputs include the user's original question, intent tags, and retrieved knowledge fragments; output is a structured natural language format answer. Different dedicated integral question-and-answer prompts are constructed for different intents to help the large-scale model better understand the meaning of the user's question. An enhanced retrieval submodule combines user questions, user intents, the knowledge base, and user points for semantic retrieval, using a vectorized search engine to improve recall. It returns the Top-N most relevant knowledge entries or FAQs (Frequently Asked Questions). Personalized sorting is performed based on user historical behavior to obtain comprehensive information for integral question-and-answer sessions.

[0131] like Figure 8 As shown, the entire process of user Q&A is as follows: Step S7.1: User asks a question. Receive voice or text input from the user; Step S7.2: Speech Recognition. Convert the speech signal into readable text. Speech recognition can be performed using deep learning speech recognition models, such as wav2vec2, etc. Step S7.3: Intelligent Error Correction. Semantic error correction and spelling correction are performed on the recognized text, outputting standardized, error-free text. Step S7.4: Intent Recognition. Identify user intent, return results for intents that do not conform to the scoring criteria, extract key information based on the intent for further reasoning; a knowledge base provides knowledge support for intent understanding and result generation. Step S7.5: Structural Reasoning. Generate a structured answer by combining user intent, key information, user questions, knowledge base, and other relevant data.

[0132] Step S7.6: Output Results. Present the reasoning results to business personnel in text or voice format.

[0133] It is understood that the various method embodiments mentioned above in this disclosure can be combined with each other to form combined embodiments without violating the principle and logic. Due to space limitations, this disclosure will not elaborate further. Those skilled in the art will understand that in the above methods of specific implementation, the specific execution order of each step should be determined by its function and possible internal logic.

[0134] In addition, this disclosure also provides a system for generating and operating business points, electronic devices, and computer-readable storage media, all of which can be used to implement any of the methods for generating and operating business points provided in this disclosure. The corresponding technical solutions and descriptions are described in the relevant section on methods and will not be repeated here.

[0135] Figure 9 This is a block diagram of a business points generation and operation system provided in an embodiment of the present disclosure.

[0136] Reference Figure 9 This disclosure provides a business points generation and operation system 900, which includes a business data acquisition module 901, a points data acquisition module 902, and an audit module 903.

[0137] The business data acquisition module 901 is used to: collect the business data required for points from multi-source heterogeneous systems, and obtain the basic data for generating points based on the collected business data. The business data required for points includes data from tables related to business income and expenditure and performance evaluation.

[0138] The points data acquisition module 902 is used to obtain points data for each level based on the basic data according to the preset points rules for each level.

[0139] The audit module 903 is used to: audit the point data of each level according to the preset point audit rules, and obtain the audit results. If the audit result is that the point data fails the audit, the first alarm message is triggered and the relevant responsible person is notified to review it so that the point data of each level passes the audit, and the audited point data of each level is obtained, which is used as the basis for calculating employee salaries.

[0140] Figure 10 This is a block diagram of an electronic device provided in an embodiment of the present disclosure.

[0141] Reference Figure 10 This disclosure provides an electronic device comprising: at least one processor 1001; at least one memory 1002; and one or more I / O interfaces 1003 connected between the processor 1001 and the memory 1002; wherein the memory 1002 stores one or more computer programs executable by the at least one processor 1001, the one or more computer programs being executed by the at least one processor 1001 to enable the at least one processor 1001 to perform the above-described method for generating and operating business points.

[0142] This disclosure also provides a computer-readable storage medium storing a computer program thereon, wherein the computer program, when executed by a processor, implements the above-described method for generating and operating business points. The computer-readable storage medium may be volatile or non-volatile.

[0143] This disclosure also provides a computer program product, including computer-readable code, or a non-volatile computer-readable storage medium carrying computer-readable code, wherein when the computer-readable code is run in a processor of an electronic device, the processor in the electronic device executes the above-described method for generating and operating business points.

[0144] Those skilled in the art will understand that all or some of the steps, systems, and apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software can be distributed on a computer-readable storage medium, which may include computer storage media (or non-transitory media) and communication media (or transient media).

[0145] As is known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information, such as computer-readable program instructions, data structures, program modules, or other data. Computer storage media includes, but is not limited to, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), static random access memory (SRAM), flash memory or other memory technologies, portable compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, it is known to those skilled in the art that communication media typically contain computer-readable program instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0146] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0147] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute 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 a remote computer, the remote computer may 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 may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0148] The computer program product described herein can be implemented specifically through hardware, software, or a combination thereof. In one alternative embodiment, the computer program product is specifically embodied in a computer storage medium; in another alternative embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0149] Various aspects of this disclosure are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer-readable program instructions.

[0150] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processor of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner; thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0151] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0152] 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 the present disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the drawings. For example, two consecutive 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, may be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0153] Example embodiments have been disclosed herein, and while specific terminology has been used, it is for illustrative purposes only and should be construed as such, and is not intended to be limiting. In some instances, it will be apparent to those skilled in the art that features, characteristics, and / or elements described in connection with particular embodiments may be used alone, or in combination with features, characteristics, and / or elements described in connection with other embodiments, unless otherwise expressly indicated. Therefore, those skilled in the art will understand that various changes in form and detail may be made without departing from the scope of this disclosure as set forth by the appended claims.

Claims

1. A method for generating and operating business points, characterized in that, include: S10: Collect the business data required for integration from the multi-source heterogeneous system, and obtain the basic data for generating integration based on the collected business data; The business data required for the points system includes data from tables related to business revenue and expenditure and performance evaluation. S20, according to the preset integration rules for each level, obtain the integration data for each level based on the basic data; S30, the points data of each level are audited according to the preset points audit rules, and the audit results are obtained; if the audit result is that the points data fails the audit, the first alarm information is triggered and the relevant responsible persons are notified to review it so that the points data of each level passes the audit, and the audited points data of each level is obtained, so as to use the audited points data of each level as the basis for calculating employee salaries.

2. The method for generating and operating business points according to claim 1, characterized in that, S10 includes: S101, The collected business data is standardized by field regular expression mapping to obtain business data with a uniform format; S102, extract the processing time, device MAC address, and service record from the standardized business data, and construct a user-business association model based on the processing time, device MAC address, and service record; S103, using the user-business association model to match offline businesses with online accounts, complete the association relationship between offline businesses and online accounts, obtain the completed data, and use the completed data as the basic data for generating points.

3. The method for generating and operating business points according to claim 1, characterized in that, S20 further includes: in response to changes in the preset scoring rules of each level, obtaining the changed scoring rules of each level, so as to use the changed scoring rules of each level to obtain the scoring data of each level based on the basic data; and recording the modifier, the modification time and the content of the change.

4. The method for generating and operating business points according to claim 1, characterized in that, S30 includes: S301, preprocess and extract features from the integral data at each level to obtain the extracted features; S302, according to the preset integral audit rules, use the isolated forest model and Kmeans clustering model to perform anomaly detection on the extracted features; the isolated forest model outputs the anomaly score of the integral numbers in the integral data of each level that are abnormal, and the Kmeans clustering model outputs the anomaly score of the behavior in the integral data of each level and the business attribution label. S303, invoke the QWEN large model based on the Transformer architecture, combine the integral knowledge base to perform business-oriented reasoning on the business attribution tags, generate reasoning conclusions, and generate an audit report based on the anomaly scores and reasoning conclusions. The audit report includes problem distribution, anomaly causes and improvement suggestions. The integral knowledge base includes a preset integral rule library at each level, preset integral audit rules and a historical anomaly case library. S304, Determine the audit results based on the audit report.

5. The method for generating and operating business points according to claim 1, characterized in that, S30 also includes: periodically checking the completeness of the basic data and obtaining the detection results; if the detection results still do not contain the basic data when the latest time limit is reached, a second alarm message is triggered and the relevant responsible persons are notified so that the basic data is complete.

6. The method for generating and operating business points according to claim 1, characterized in that, Also includes: The reporting system displays the audited point data at each level according to access permissions, generates point reports, and pushes them to relevant personnel.

7. The method for generating and operating business points according to claim 1, characterized in that, Also includes: We provide customer service to employees through speech recognition, intelligent error correction, intent recognition, knowledge base retrieval, and large-scale models. The customer service includes points inquiry, rule explanation, anomaly feedback, activity consultation, and appeals.

8. The method for generating and operating business points according to claim 7, characterized in that, Also includes: An abnormal behavior feature database is constructed. During the process of obtaining the points data at each level based on the basic data, the abnormal behavior is intercepted in real time by combining the real-time rule engine and audit report, and risk control logs are generated. The abnormal behavior includes short-term high-frequency fake orders, points operation by multiple accounts on the same IP, and abnormal cross-regional business processing.

9. A system for generating and operating business points, characterized in that, include: The business data acquisition module is used to: collect the business data required for points from multi-source heterogeneous systems, and obtain the basic data for generating points based on the collected business data; the business data required for points includes data from tables related to business income and expenditure and performance evaluation. The points data acquisition module is used to: obtain points data for each level based on the basic data according to the preset points rules for each level; The audit module is used to: audit the points data of each level according to the preset points audit rules, and obtain the audit results; if the audit result is that the points data fails the audit, the first alarm message is triggered and the relevant responsible persons are notified to review it so that the points data of each level passes the audit, and the audited points data of each level is obtained, so as to use the audited points data of each level as the basis for calculating employee salaries.

10. An electronic device, characterized in that, include: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores one or more computer programs that can be executed by the at least one processor, the one or more computer programs being executed by the at least one processor to enable the at least one processor to perform the method for generating and operating business points as described in any one of claims 1-8.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method for generating and operating business points as described in any one of claims 1-8.

12. A computer program product, characterized in that, Includes computer-readable code, or a non-volatile computer-readable storage medium carrying computer-readable code, wherein when the computer-readable code is run in a processor of an electronic device, the processor in the electronic device performs a method for generating and operating business points as described in any one of claims 1-8.

Images