A data evaluation method and device, electronic equipment and storage medium

Abstract

This application discloses a data evaluation method, apparatus, electronic device, and storage medium. The method includes: acquiring data to be evaluated associated with a target object; converting the data to be evaluated into structured data corresponding to the target object, wherein the structured data includes at least one field and field values ​​corresponding to the fields; searching a preset rule base based on the structured data corresponding to the target object to determine a set of rule items matching the target object; performing evaluation processing on at least one dimension based on the structured data and the set of rule items to determine at least one evaluation result for the target object; and generating an evaluation report for the target object according to a preset evaluation report template based on the at least one evaluation result for the target object. Therefore, an evaluation report for the target object can be generated based on the data to be evaluated associated with the target object, improving the efficiency and accuracy of data evaluation.

Description

Technical Field

[0001] This application relates to the field of data processing technology, specifically to a data evaluation method, apparatus, electronic device, and storage medium. Background Technology

[0002] In many business scenarios, it is often necessary to evaluate target objects (such as enterprises, individuals, projects, equipment, and transactions) to provide a basis for subsequent decision-making, risk control, or performance management. For example, financial institutions need to conduct credit assessments of borrowing companies before approval, medical systems need to conduct risk assessments of patient health data, and operations and maintenance departments need to conduct performance assessments of equipment operating status. The accuracy of the evaluation results affects the quality of decision-making and business efficiency. However, the methods for evaluating target objects often require manual calculation or judgment of each indicator according to pre-set rules and parameters (such as thresholds, weights, and judgment conditions) to arrive at an evaluation conclusion. This process is labor-intensive, inefficient, and prone to human error. Summary of the Invention

[0003] This application provides a data evaluation method, apparatus, electronic device, and storage medium, which can more comprehensively and objectively reflect the actual situation of the target object, thereby generating an evaluation report of the target object and improving the efficiency and accuracy of data evaluation.

[0004] In a first aspect, embodiments of this application provide a data evaluation method, the method comprising: Obtain the data to be evaluated associated with the target object, and transform the data to be evaluated into structured data corresponding to the target object. The structured data includes at least one field and the field value corresponding to the field. Based on the structured data corresponding to the target object, a search is performed in a preset rule base to determine the set of rule items that match the target object; Based on the structured data and the set of rule items, perform evaluation processing on at least one dimension to determine at least one evaluation result for the target object; An evaluation report for the target object is generated according to a preset evaluation report template based on at least one evaluation result of the target object.

[0005] Secondly, embodiments of this application provide a data evaluation apparatus, the apparatus comprising: The acquisition module is used to acquire the data to be evaluated associated with the target object, and to convert the data to be evaluated into structured data corresponding to the target object. The structured data includes at least one field and the field value corresponding to the field. The retrieval module is used to search in a preset rule base based on the structured data corresponding to the target object to determine the set of rule items that match the target object; An evaluation module is used to perform evaluation processing on at least one dimension based on the structured data and the set of rule items, and to determine at least one evaluation result for the target object; The output module is used to generate an evaluation report for the target object according to a preset evaluation report template based on at least one evaluation result of the target object.

[0006] Thirdly, embodiments of this application also provide an electronic device, including a memory storing multiple instructions; a processor loads instructions from the memory to execute the steps of any of the data evaluation methods provided in embodiments of this application.

[0007] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a plurality of instructions adapted for loading by a processor to perform the steps of any of the data evaluation methods provided in embodiments of this application.

[0008] Fifthly, embodiments of this application also provide a computer program product, including a computer program or instructions, which, when executed by a processor, implement the steps in any of the data evaluation methods provided in embodiments of this application.

[0009] The solution adopted in this application embodiment retrieves and matches structured data in a preset rule base, which can obtain a set of rule items that match the target object, making the subsequent evaluation process more targeted and objective, and avoiding the bias that may occur when manual selection of rule items based on experience; multi-dimensional evaluation processing is performed based on structured data and the set of rule items, thereby generating an evaluation report of the target object, which can improve the efficiency and accuracy of data evaluation. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is an application scenario diagram of the data evaluation method provided in some embodiments of this application; Figure 2 This is a schematic flowchart of a data evaluation method provided in some embodiments of this application; Figure 3 This is yet another schematic diagram of a data evaluation process provided in some embodiments of this application; Figure 4 These are exemplary schematic diagrams of data evaluation apparatuses provided in some embodiments of this application; Figure 5 These are exemplary schematic diagrams of electronic devices provided in some embodiments of this application. Detailed Implementation

[0012] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. At the same time, in the description of the embodiments of this application, the terms "first," "second," etc., are only used to distinguish descriptions and should not be construed as indicating or implying relative importance. Thus, features defined with "first" and "second" may explicitly or implicitly include one or more features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0013] This application provides a data evaluation method, apparatus, electronic device, and computer-readable storage medium.

[0014] Specifically, this embodiment will be described from the perspective of a data evaluation device, which can be integrated into an electronic device. That is, the data evaluation method of this application embodiment can be executed by an electronic device. Optionally, the electronic device may include a terminal device. The terminal device may be a mobile phone, tablet computer, smart Bluetooth device, laptop computer, game console, or personal computer (PC), etc.

[0015] Please see Figure 1 ,like Figure 1 As shown, the data evaluation method provided in this application embodiment can be applied to, for example... Figure 1 The data evaluation system may include terminal devices and servers. The terminal devices may be devices with receiving and transmitting hardware, specifically devices capable of performing bidirectional communication over a bidirectional communication link. The terminal devices and servers can communicate bidirectionally via a network.

[0016] Optionally, the server can be a standalone server, or a server network or server cluster, including but not limited to computers, network hosts, single network servers, multiple network server sets, or cloud servers composed of multiple servers. Cloud servers consist of a large number of computers or network servers based on cloud computing.

[0017] The following detailed description is provided in conjunction with the accompanying drawings. In this embodiment, the execution subject is a terminal device as an example. It should be noted that the order of description in the following embodiments is not intended to limit the preferred order of the embodiments. Although a logical order is shown in the flowcharts, in some cases, the steps shown or described may be performed in a different order than that shown in the accompanying drawings.

[0018] This application provides a data evaluation method. Please refer to the embodiments provided. Figure 2 , Figure 2 The specific process of the data evaluation method provided in this application embodiment includes the following steps 210 to 240: Step 210: Obtain the data to be evaluated associated with the target object, and transform the data to be evaluated into structured data corresponding to the target object. The structured data includes at least one field and the field value corresponding to the field.

[0019] The target object refers to the object of the assessment and processing. The target object can be an individual, enterprise, other organization, project, or virtual entity, etc. In practical application scenarios, the target object can be a corporate client applying for credit rating, a user undergoing health assessment, a fixed asset to be valued, or a financial product to be analyzed for investment risk, etc. This application embodiment does not limit this.

[0020] The data to be evaluated refers to the original set of data associated with the target object. In some embodiments, the data to be evaluated associated with the target object can be obtained through manual input or a database.

[0021] Structured data refers to data obtained by processing unstructured or semi-structured data to be evaluated. Structured data contains at least one field and its corresponding value. The field describes the attribute of the target object in a specific dimension, such as the attribute name corresponding to attributes like "enterprise registration number," "document author," or "sensor identifier." The "field value" is the value of that attribute in the current state of the target object, such as "unified social credit code" as XXXX, "username" as "Zhang San," or "temperature sensor number 001." A piece of structured data typically consists of a set of multiple fields and field values.

[0022] The following explanation uses a specific application scenario. In some embodiments, the target objects can be configured to be multiple, for example, all employees within an enterprise who need to be audited. Correspondingly, the data to be evaluated can be a spreadsheet submitted by a specific department of the enterprise (such as the human resources department or the finance department), such as a "Monthly Salary Sheet". This data to be evaluated can be in spreadsheet format (e.g., .xlsx or .csv), which can contain salary information for multiple target objects (such as employees) in a certain month.

[0023] In some embodiments, each row of the spreadsheet can be read, and the column names in the header (such as "Employee ID", "Employee Name", "Basic Salary", "Performance Bonus", "Social Security Deductions", "Housing Provident Fund Deductions", "Net Salary") can be mapped to preset standard field names. The specific content of the cell corresponding to the standard field name in each row is used as the field value of the corresponding field, thus obtaining the structured data corresponding to the target object. In an optional embodiment, there can be multiple target objects, and each row of records can be regarded as the structured data of an independent target object (i.e., each employee). Thus, a list consisting of a set of structured data is obtained, where each structured data contains at least one field and its corresponding field value. For example, for the target object with employee ID "E1", its structured data includes the field value of "Basic Salary" as "XX yuan", the field value of "Performance Bonus" as "XX yuan", and the field value of "Net Salary" as "XX yuan".

[0024] Step 220: Based on the structured data corresponding to the target object, search in the preset rule base to determine the set of rule items that match the target object.

[0025] A pre-built rule base is a collection of knowledge.

[0026] In some embodiments, the preset rule base may contain several rule entries, each rule entry containing at least one conditional expression and one decision logic or execution action. The conditional expression may be composed of a logical expression used to judge fields and field values ​​in structured data, such as "field A equals the corresponding threshold a", "field B is greater than the corresponding threshold b", or "field C belongs to a specific set". The decision logic or execution action refers to one or more parameter values ​​output when the logical expression is satisfied, such as weight value, grade coefficient, risk score, processing identifier, Boolean value indicating whether it passes or fails, etc.

[0027] In some embodiments, the structured data corresponding to multiple target objects can be represented in a standardized JSON format. For example, the structured data for employee Zhang San may include fields such as "Employee Name" as "Zhang San", "Department" as "Technical Department", "Overtime Hours" as "20 hours", "Performance Rating" as "S", and "Salary" as "15,000 yuan".

[0028] In some embodiments, for the current target object (e.g., a specific employee), relevant rule description information can be retrieved from a preset rule base, and executable rule items can be extracted from the rule description information using a large language model. Examples include "the overtime multiplier is 1.5," "the maximum overtime hours is 36 hours per month," and "the bonus coefficient corresponding to performance level S is 1.2."

[0029] In some embodiments, based on the structured data corresponding to the target object, a search is performed in a preset rule base to determine a set of rule items matching the target object, including: Based on the fields in the structured data and the preset mapping relationships, the business scenario tags corresponding to the fields are obtained. The preset mapping relationships include the preset business scenario tags corresponding to each preset field. Based on the business scenario tags, retrieve the rule description information associated with the business scenario tags from the preset rule base; The executable rule items are extracted from the rule description information by a large language model, resulting in a set of rule items.

[0030] In some embodiments, the preset rule library may contain various rule description information (e.g., policy clauses) related to assessment (e.g., salary audit), such as "Overtime Calculation Rules", "Performance Appraisal Methods", "Salary Calculation Standards", etc. For example, the preset rule library stores the reference vector corresponding to each rule description information, as well as the business scenario tag to which the rule description information belongs, such as tags like "Overtime Calculation Scenario", "Performance Appraisal Scenario", or "Salary Compliance Scenario".

[0031] In some embodiments, rule description information matching the business scenario tag corresponding to the target object can be retrieved from a preset rule base based on the structured data of the target object, thereby obtaining a set of rule items.

[0032] In some embodiments, a preset mapping table can be established in advance to map field names (e.g., "overtime hours") to preset business scenario labels (e.g., "overtime calculation scenario").

[0033] In some embodiments, query vectors for retrieval can be generated based on the business scenario tags corresponding to each currently identified field. For example: key information of the target object can be extracted from structured data; through a preset mapping relationship, scenario-related query text can be obtained, such as "employee Zhang San, technical department, overtime calculation scenario, performance appraisal scenario, salary compliance scenario"; the query text is embedded using an embedding model to obtain the corresponding query vector; and based on this query vector, a similarity search is performed in a preset rule base to obtain the reference vector with the highest matching degree. The rule description information corresponding to the reference vector with the highest matching degree is used as several rule description information matching the target object.

[0034] In some embodiments, a corresponding query vector can be generated for the structured data of each target object, and retrieval can be performed based on the query vector of each target object to avoid retrieval bias caused by mixing multiple records. In some embodiments, for rules with the same scenario (e.g., rule description information applicable to overtime work for all employees), the initial retrieval results can be cached to avoid repeated retrieval, thereby improving overall efficiency.

[0035] In some embodiments, once matching rule description information is obtained, such as the clause description text in the "Overtime Calculation Rules," a large language model can be invoked to extract at least one executable rule item from the clause description text, forming a set of rule items for the current target object and its business scenario label (overtime calculation scenario). For example, if the clause description text states "Overtime on weekdays is calculated at 1.5 times the salary, and overtime on rest days is calculated at 2 times the salary," the large language model will parse it into structured rule items: the multiplier for overtime on weekdays is 1.5, and the multiplier for overtime on rest days is 2.0.

[0036] Step 230: Perform evaluation processing on at least one dimension based on the structured data and the set of rule items to determine at least one evaluation result for the target object.

[0037] Each dimension can correspond to an independent evaluation metric. For example, in a payroll audit scenario, dimensions could include compliance, accuracy, or reasonableness. Each dimension can also correspond to an independent evaluation logic.

[0038] The evaluation result refers to the result obtained after performing evaluation processing on a specific dimension. In some embodiments, the evaluation result may be presented in the form of, but is not limited to, numerical scores (e.g., 85 points), rating labels (e.g., "Grade A", "Compliant", "High Risk"), Boolean values ​​(e.g., "Pass", "Fail"), or descriptive text (e.g., "Overtime hours exceeded the legal limit of 6 hours"). For the same target object, an evaluation result for one dimension may be output, or evaluation results for multiple dimensions may be output to reflect the status of the target object in different aspects.

[0039] In some embodiments, the evaluation process for at least one dimension includes at least one of consistency evaluation, compliance evaluation, accuracy evaluation, and logical evaluation.

[0040] Consistency assessment refers to verifying whether the values ​​of the same or related fields in structured data match across different data sources and records. For example, in a payroll audit scenario, consistency assessment can compare the "net salary" field in the "Monthly Payroll Table" with the "actual transfer amount" field in the bank's payroll file. If the two values ​​match, the assessment passes; otherwise, it is considered an anomaly.

[0041] Compliance assessment refers to verifying whether field values ​​in structured data meet preset standards. For example, a compliance assessment can determine whether a target's "overtime hours" exceed the stipulated monthly limit of 36 hours, or whether their "probationary period salary" is less than 80% of their regular salary.

[0042] Accuracy assessment refers to verifying the correctness of numerical calculations, formulas, or data transformations in structured data. For example, accuracy assessment can involve recalculating "basic salary" + "performance bonus" + "overtime pay" minus "absence deductions" in a payroll table, and comparing the result with the gross salary in the monthly payroll table to determine if there are any errors.

[0043] Logical evaluation refers to checking whether the logical relationships between fields in structured data are reasonable and whether they conform to conventional business semantics or causal relationships. For example, logical evaluation may include: if the performance bonus of an employee with a "performance grade" of "D" is higher than that of an employee with a "performance grade" of "A", then a logical anomaly is determined.

[0044] In some embodiments, for a specific dimension, the evaluation result corresponding to the evaluation process may include anomaly information, such as anomaly type and corresponding anomaly description information. Anomaly type refers to the result of classifying abnormal events during the evaluation process, such as "overtime anomaly," "data inconsistency anomaly," "calculation error anomaly," "logical contradiction anomaly," or "compliance violation anomaly." Anomaly description information is information related to a specific anomaly type, such as the severity, deviation magnitude or value, and other textual information. For example, the evaluation result of a compliance assessment may include, when "overtime hours" exceed the legal limit, determining the anomaly type as "overtime," with corresponding anomaly description information such as "exceeding the limit by 5 hours" or "exceeding the limit by 13.9%." The evaluation result of an accuracy assessment may include, when an error is detected in the calculation of "gross pay," determining the anomaly type as "calculation error," with corresponding anomaly description information such as "absolute difference of 200 yuan" or "relative error rate of 2.5%." The results of the consistency assessment can include, for example, classifying an anomaly as "data inconsistency" when a discrepancy is detected between "actual salary" and bank statements, with the corresponding anomaly description being "the difference is -150 yuan". The results of the logical reasoning assessment can include, for example, classifying an anomaly as "logical contradiction" when a contradiction is detected between "seniority-based salary" and "date of employment," with the corresponding anomaly description being "seniority-based salary should be 0, but is actually 300 yuan".

[0045] Step 240: Generate an evaluation report for the target object based on at least one evaluation result of the target object and according to a preset evaluation report template.

[0046] A pre-designed assessment report template refers to a pre-designed document framework. For example, a pre-designed assessment report template might contain two parts: fixed content and variable content. Fixed content refers to the parts that do not change with the assessment results, such as titles, labels, fixed headings for each chapter, table header row definitions, and explanatory footnotes. Variable content refers to the parts of the assessment report template that are to be filled in, such as variable names enclosed in double curly braces, indicating that the corresponding assessment results will be dynamically filled into those positions during the report generation phase.

[0047] An evaluation report is the final document generated and output. For example, an evaluation report may include attribute information of the target object, key information in structured data, evaluation results of each dimension, and opinions or recommendations based on these results.

[0048] In some embodiments, when there are multiple target objects, the evaluation report may include the evaluation results of multiple target objects.

[0049] In some embodiments, the evaluation report may be in the form of a document file, a Hypertext Markup Language page, or a text summary, to facilitate subsequent manual review, archiving, or reporting.

[0050] In some embodiments of this application, by uniformly converting data to be evaluated in different formats into structured data containing clearly defined fields and field values, manual data entry and format conversion can be avoided, improving data processing efficiency and preventing data errors caused by manual processing. Secondly, by searching and matching the structured data in a preset rule base, a set of rule items matching the target object can be obtained, making subsequent evaluations more targeted and objective, avoiding biases that may arise from manual selection of rule items. Performing multi-dimensional evaluation processing based on the structured data and the set of rule items, and then generating an evaluation report for the target object, can improve the efficiency and accuracy of data evaluation.

[0051] In some embodiments, evaluation processing is performed on at least one dimension based on structured data and a set of rule items to determine at least one evaluation result for the target object, including: Based on the dimension to which each evaluation process belongs, determine the set of sub-rule items required to complete each evaluation process from the set of rule items; Based on the set of sub-rule items required to complete each evaluation process, the structured data is evaluated accordingly, and the evaluation results corresponding to each evaluation process are determined.

[0052] In some embodiments, the rule item set may include various rule items applicable to different evaluation dimensions. For example, during a payroll audit, the rule item set obtained by matching structured data of a target object from a preset rule base may simultaneously include the following rule items: "Overtime multiplier on weekdays is 1.5 times", "Overtime multiplier on rest days is 2 times", "Monthly overtime hours limit is 36 hours"; "S-level performance coefficient is 1.2", "A-level performance coefficient is 1.0"; "Tax threshold is 5,000 yuan", "Social security individual deduction rate is 8%", etc.

[0053] Different evaluation processes correspond to different rule items, and the correspondence can be determined based on prior knowledge or historical data.

[0054] For example, if the current assessment process includes compliance assessment and accuracy assessment, then for the compliance assessment, the corresponding relationship is identified as "Compliance - Overtime Hours". The rule items related to overtime hours are searched from the rule item set, and the rule item "Monthly overtime hours capped at 36 hours" is extracted to form the sub-rule item set required for the compliance assessment. For the accuracy assessment, the corresponding relationship is identified as "Accuracy - Overtime Pay Calculation". The rule items related to overtime pay calculation are searched from the rule item set, and the rule items "Weekday overtime multiplier is 1.5 times" and "Rest day overtime multiplier is 2 times" are extracted to form the sub-rule item set required for the accuracy assessment.

[0055] In some embodiments, after extracting the set of sub-rule items, corresponding evaluation processing is performed on the structured data of the target object according to each set of sub-rule items to determine the evaluation result corresponding to each evaluation processing. Taking compliance evaluation as an example, the set of sub-rule items required to complete the compliance evaluation is obtained, such as "the maximum overtime hours per month is 36 hours". The value of the "actual overtime hours in the current month" field of the target object is read from the structured data as 20 hours. A comparison operation is performed: 20 hours are compared with 36 hours, and the comparison result shows that 20 hours is less than 36 hours. Therefore, the evaluation result for this dimension is "compliant". If the value of the "actual overtime hours in the current month" field of the target object is 40 hours, the evaluation result is "violation", the exception type is "overtime exception", and the exception description information is "exceeded by 4 hours". For example, for accuracy assessment, the set of sub-rules required to complete the accuracy assessment is obtained, such as "the overtime multiplier on weekdays is 1.5 times" and "the overtime multiplier on rest days is 2 times". From structured data, the "basic salary" is read as 8000 yuan, the "overtime hours on weekdays" as 15 hours, the "overtime hours on rest days" as 5 hours, and the "actual overtime pay" as 1500 yuan. The theoretical overtime pay is calculated according to a preset formula, such as: the overtime pay on weekdays equals the basic salary divided by the number of working days in the month (21.75 days), then divided by 8 hours per day, multiplied by 15 hours, and then multiplied by 1.5 times; the overtime pay on rest days can be obtained using a similar method. The theoretical overtime pay is calculated and compared with the value of 1500 yuan in the "Actual Overtime Pay" field. If they match, the evaluation result is "Accurate"; if they do not match, the evaluation result is "Inaccurate" and the exception type is "Calculation Error Exception". The exception description information is "Difference amount is negative 200 yuan" or "Relative error rate is 1.3%", etc.

[0056] In some embodiments, the structured data is subjected to corresponding evaluation processing based on the set of sub-rule items required to complete each evaluation process, and the evaluation result corresponding to each evaluation process is determined, including: Based on the set of sub-rule items corresponding to the evaluation process, rule evaluation is performed on the structured data to obtain the first sub-evaluation result corresponding to the evaluation process; and / or, By using a large language model to perform semantic evaluation on structured data based on the set of sub-rule items corresponding to the evaluation process, the second sub-evaluation result corresponding to the evaluation process is obtained. Based on the results of the first sub-evaluation and / or the second sub-evaluation, determine the evaluation result corresponding to the evaluation process.

[0057] Rule-based evaluation can include, but is not limited to, numerical comparison operations based on logical expressions and mathematical formula calculations. For example, in accuracy evaluation, the gross salary can be recalculated according to a preset mathematical formula and compared with the actual salary in the structured data to obtain a first sub-evaluation result of "accurate" or "inaccurate".

[0058] In some embodiments, a large language model can also be used to perform semantic evaluation on structured data based on the set of sub-rule items corresponding to the evaluation process, resulting in a second sub-evaluation result. Semantic evaluation is suitable for scenarios where rules are fuzzy, semantic understanding is required, or text reasoning is involved. For example, the input to the large language model may include the set of sub-rule items corresponding to the evaluation process, and the output may include the second sub-evaluation result.

[0059] For example, in a compliance assessment, the set of sub-rule items could include the fuzzy rule item "Employees with excellent performance reviews for two consecutive months may receive an additional bonus, the specific amount of which shall be determined by the department head based on the actual situation, but shall not exceed 20% of their basic salary." The structured data includes: excellent performance reviews for two consecutive months, a basic salary of 8000 yuan, and an actual additional bonus of 2000 yuan. Since it is impossible to determine whether 2000 yuan is reasonable under the "actual situation determination" through rule judgment alone, a large language model can be used to combine the semantics and contextual information of the rule description to reason and output the second sub-assessment result.

[0060] In some embodiments, the final evaluation result can be determined based on the results of the first sub-evaluation and / or the second sub-evaluation. For example, when only rule evaluation is performed, the result of the first sub-evaluation is directly used as the final evaluation result. When only semantic evaluation of the large language model is performed, the result of the second sub-evaluation is directly used as the final evaluation result. When both are performed, a fusion approach can be used to determine the final evaluation result. For example, if the result of the first sub-evaluation is "compliant" and has a high confidence level, the result of the first sub-evaluation is directly used as the final evaluation result; if the result of the first sub-evaluation is "non-compliant" or has a low confidence level, the large language model is invoked for semantic evaluation, and the result of the second sub-evaluation is used for correction. Alternatively, the final evaluation result can be output according to a preset priority (e.g., based on the result of the large language model, or based on the result of the deterministic rule).

[0061] By combining the determinism of rule evaluation with the flexibility of large language model evaluation, we can improve the comprehensiveness and accuracy of evaluation and reduce the risk of misjudgment and omission when dealing with complex evaluation scenarios that include both numerical and semantically ambiguous rule items.

[0062] In some embodiments, determining the evaluation result corresponding to the evaluation process based on the first sub-evaluation result and / or the second sub-evaluation result includes: Based on the distribution characteristics of the fuzzy rule items in the sub-rule item set, determine the first weight corresponding to the first sub-evaluation result and the second weight corresponding to the second sub-evaluation result; The first sub-evaluation result and the second sub-evaluation result are weighted and fused according to the first weight and the second weight to obtain the evaluation result corresponding to the evaluation process.

[0063] In some embodiments, the set of sub-rule items may include precise rule items and fuzzy rule items. Precise rule items may be rule items containing explicit logical expressions. Fuzzy rule items refer to rule items described in natural language and whose semantics are ambiguous.

[0064] Distribution characteristics refer to the distribution of fuzzy rule terms within the set of sub-rule terms. Specifically, distribution characteristics can be expressed as: whether the set of sub-rule terms contains at least one fuzzy rule term, the proportion of fuzzy rule terms to the total number of sub-rule terms, the importance of the fields involved in the fuzzy rule terms, or the complexity level of the fuzzy rule terms, etc. In some embodiments, the distribution characteristic is the presence of at least one fuzzy rule term. When the set of sub-rule terms contains at least one fuzzy rule term, the second sub-evaluation result (output of the large language model) can be assigned a higher weight, such as 0.8, and the first sub-evaluation result (output of the rule engine) can be assigned a lower weight, such as 0.2. Conversely, if the set of sub-rule terms does not contain any fuzzy rule terms, the first sub-evaluation result can be assigned a higher weight, such as 0.8, and the second sub-evaluation result can be assigned a lower weight, such as 0.2. Alternatively, in some cases, the large language model does not participate in the evaluation.

[0065] In some embodiments, the first and second sub-evaluation results can be weighted and fused. The weighting and fusion method can depend on the data type of the evaluation results. For example, if the first and second sub-evaluation results are numerical, such as the first sub-evaluation result being 85 points and the second sub-evaluation result being 95 points, then the weighted fusion result can be 85 multiplied by the first weight + 95 multiplied by the second weight. Alternatively, if the first and second sub-evaluation results are Boolean or categorical (e.g., pass or fail), then soft voting or a threshold method can be used. For example, pass can be mapped to 1 and fail to 0, a weighted sum can be calculated, and then the final evaluation result can be determined based on a preset threshold (e.g., 0.5). In a payroll audit scenario, to facilitate the generation of evaluation reports, the first and second sub-evaluation results can be converted into levels of "compliant," "suspected violation," and "violation," and a weighted score can be calculated based on the weights. For example, compliance is scored as 1 point, suspected violation as 0.5 points, and violation as 0 points. When the weighted score is greater than or equal to 0.7, the final assessment result is "compliant"; when it is between 0.3 and 0.7, the final assessment result is "suspected violation"; when it is less than 0.3, the final assessment result is "violation", etc.

[0066] By using weighted processing, leveraging the efficiency and determinism of the rule engine, and the flexibility of the large language model in handling fuzzy rule items, the relative importance of the two is automatically adjusted based on the distribution characteristics of the fuzzy rule items, thus avoiding misjudgments that may occur under different sets of rule items with fixed weights.

[0067] Please see Figure 3 , Figure 3 Here is a specific flowchart of another data evaluation method provided in the embodiments of this application, as follows: Steps 310 to 320: In some embodiments, an evaluation report for the target object is generated according to a preset evaluation report template based on at least one evaluation result of the target object, including: Step 310: Integrate the evaluation results corresponding to each evaluation treatment of the target object to determine the comprehensive evaluation result corresponding to the target object; Step 320: Based on the comprehensive evaluation results corresponding to the target object, generate an evaluation report for the target object according to the preset evaluation report template.

[0068] In some embodiments, the comprehensive evaluation result includes clustered anomaly information corresponding to the target object, and / or a comprehensive evaluation index corresponding to the target object. The comprehensive evaluation result for the target object is determined by integrating the evaluation results corresponding to each evaluation process, including: When the evaluation results contain corresponding anomaly information, semantic clustering is performed based on the anomaly information corresponding to each evaluation result to obtain the clustered anomaly information corresponding to the target object; and / or, Obtain the basic assessment score and the impact coefficient corresponding to each assessment result for the target object; Based on the basic evaluation score corresponding to each evaluation result and the influence coefficient corresponding to each evaluation result, the comprehensive evaluation index corresponding to the target object is obtained.

[0069] In some embodiments, the evaluation results corresponding to each evaluation process can be the evaluation results obtained by multiple audit engines performing their respective dimension evaluation processes on the same target object, such as attendance audit engine, performance audit engine, payroll compliance engine, and data consistency engine.

[0070] In some embodiments, the attendance audit engine, performance audit engine, payroll compliance engine, and data consistency engine can be composed of a rule engine and a language model. The language model can be a large language model (LLM), which can be composed of an artificial neural network with many parameters (typically billions of weights or more) and trained using self-supervised learning or semi-supervised learning.

[0071] In some embodiments, the language models in the attendance audit engine, performance audit engine, payroll compliance engine, and data consistency engine can be the same or different language models.

[0072] In some embodiments, the various assessment results may differ. For example, the number and types of anomalies included in the various assessment results may differ. For instance, the output of the attendance audit engine includes an assessment result of "monthly cumulative overtime hours of 45 hours, exceeding the policy limit of 36 hours," and the risk level of this assessment result: medium risk, with a corresponding basic risk score of 2 points; simultaneously, the output of the performance audit engine includes an assessment result of "performance level S corresponding to a coefficient of 2.2, exceeding the policy limit of 1.5 to 2.0," and the risk level of this assessment result: medium risk, with a corresponding basic risk score of 2 points.

[0073] In some embodiments, after obtaining the set of evaluation results described above, an aggregation operation is performed. For example, the identification information of the target object, such as the employee ID, is used as the index for the aggregation operation. The evaluation results for that employee ID are extracted from all the evaluation result records, thereby filtering out all evaluation result records belonging to the same target object and forming a candidate record set to be merged.

[0074] In some embodiments, each anomaly description in the candidate record set can be converted into a feature vector. For example, the embedding layer in a large language model can be invoked to map the anomaly description of natural language text into a high-dimensional vector representation; alternatively, a general pre-trained language model, such as a transformer-based bidirectional encoder representation model, can be used to extract semantic features of the anomaly description, with each anomaly description corresponding to a feature vector.

[0075] In some embodiments, the large language model can cluster the candidate record set of the target object. For example, clustering can be performed using clustering algorithms, including but not limited to density-based noisy spatial clustering, hierarchical clustering, or connected component algorithms based on similarity thresholds. Taking a connected component algorithm based on similarity thresholds as an example, a similarity threshold is preset, for example, 0.7. If the similarity between the feature vectors of two anomaly descriptions is greater than or equal to the similarity threshold, an edge is established between them. All anomaly descriptions connected by edges constitute a cluster. Each cluster corresponds to a set of semantically similar anomaly types. An anomaly description can be selected from the cluster as the anomaly description for that cluster, or the large language model can generate a summary description based on multiple anomaly descriptions within a cluster.

[0076] In some embodiments, records generated by the same audit engine for the same anomaly type due to multiple detections can be removed before performing clustering operations.

[0077] In some embodiments, to further accommodate the varying degrees of importance of different audit engines in actual business operations, the impact coefficients of different audit engines are different, and these impact coefficients can be flexibly configured according to actual circumstances. For example, the data consistency engine and the payroll compliance engine can be configured with higher impact coefficients.

[0078] In some embodiments, before performing evaluation processing on at least one dimension based on structured data and a set of rule items to determine at least one evaluation result for the target object, the method further includes: Identify abnormal rule items in the rule item set. Abnormal rule items include rule items that are inconsistent with the preset standard rule items and rule items whose parameter values ​​deviate. Generate prompt messages based on the exception rules to prompt the user to confirm or make corrections; Retrieve the updated exception rules to update the preset rule base.

[0079] In some embodiments, it is necessary to validate the acquired set of rule items in order to identify any abnormal rule items that may exist.

[0080] Preset standard rules can be derived from preset regulations, preset standards, and rules verified in historical data. For example, in an overtime calculation scenario, if a rule is extracted stating "the monthly overtime limit is 50 hours," while the preset standard rule states a legal overtime limit of 36 hours, the 50-hour limit is inconsistent with the 36-hour limit, thus identifying "the monthly overtime limit of 50 hours" as an abnormal rule. Similarly, in a performance evaluation scenario, if a rule is extracted stating "the S-level performance coefficient is 3.0," while the standard parameter range is 1.0 to 2.0, 3.0 deviates from the normal standard parameter range, indicating a deviation in the parameter value. In some embodiments, the name, numerical range, unit, and constraints of a rule can be compared with the corresponding rule entries in the preset standard rule library to identify rule items that are inconsistent or whose parameter value deviation exceeds a preset deviation threshold, thus classifying them as abnormal rule items.

[0081] In some embodiments, the prompt information can take various forms, such as voice or pop-up windows in the interactive interface. The prompt information may include the source of the abnormal rule item (which knowledge base it comes from), the extracted rule item, the preset standard parameter value or reasonable range, and suggestions. Alternatively, the prompt information can be highlighted, and parameter values ​​that exceed a preset deviation threshold can be highlighted, such as by marking them in red.

[0082] In some embodiments, after a user completes a confirmation or modification operation through an interactive interface, the updated exception rule item is obtained. For example, if a user confirms that the 50-hour overtime limit is indeed a special rule item during a certain project and uploads relevant supporting documents, then the rule item is associated with the relevant supporting documents; or if the user changes 50 hours to 36 hours, then the updated parameter value is obtained as 36 hours, the exception rule item is updated, and the updated rule item is updated to the preset rule base.

[0083] In some embodiments, based on structured data, a search is performed in a preset rule base to determine a set of rule items matching the target object, including: Based on the structured data, a search is performed in the preset rule base to obtain a set of candidate rule items that match the target object. The set of candidate rule items includes multiple candidate rule items with inconsistent content. Based on the priority of each rule item, the target rule item is selected from multiple rule items with inconsistent content to determine the set of rule items that match the target object; Based on at least one evaluation result of the target object, and following a pre-defined evaluation report template, generate an evaluation report for the target object, including: The generated evaluation report marks multiple inconsistent rule selection items and the selected target rule selection items.

[0084] In one specific implementation, because the enterprise knowledge base may contain rule descriptions from multiple sources, multiple versions, or for different scenarios, inconsistent candidate rule items may be retrieved for the same business rule. For example, for the "overtime calculation scenario," the knowledge base may contain both the "monthly overtime limit of 36 hours" stipulated in the "Attendance Management System" and the "overtime limit can be relaxed to 50 hours during the project period" stipulated in the "Exemption Notice" for a specific project period. Therefore, when searching in the preset rule base, a set of candidate rule items matching the target object will be obtained, which may contain multiple inconsistent candidate rule items. In this case, a priority can be pre-configured for each candidate rule item, and the candidate rule item with the highest priority can be selected as the target rule item based on the priority of each candidate rule item.

[0085] In one specific implementation, the evaluation report can list the source, specific content, priority, and selection status of all candidate rule items in tabular or list form. For candidate rule items that are not selected, the reason for not selecting them should be noted, such as "not adopted due to low priority" or "not adopted due to insufficient timeliness (obsolete version)". The final selected target rule item is highlighted or bolded. For example, the evaluation report can be described as follows: "This evaluation retrieved three candidate rule items for the overtime limit: Rule item 1 comes from the 'Attendance Management System', with an overtime limit of 36 hours and a general priority; Rule item 2 comes from the 'Exemption Notice', with an overtime limit of 50 hours and a special priority. According to priority, special level is higher than general level, therefore, this evaluation selected rule item 2, i.e., the overtime limit of 50 hours, as the basis for judgment."

[0086] By displaying candidate rule items and selected target rule items in the evaluation report, the basis for the evaluation results is clearly traceable. When it is necessary to verify the evaluation results, it can be traced back to the rule selection stage, thereby improving the reliability, credibility and convenience of the evaluation report.

[0087] To better understand the above scheme, the data evaluation method will be explained below with a specific embodiment.

[0088] In one implementation, users submit monthly payroll tables for auditing via an interactive interface. These tables can be in Excel or CSV format. Using regular expression matching and row / column parsing techniques, the unstructured tables are transformed into structured data, such as JSON data. The generated structured data includes key fields such as employee ID, name, department, attendance, performance, and salary.

[0089] In some embodiments, information from structured data, such as "Employee Zhang San works 20 hours of overtime on weekends," can be combined into natural language query text. This query text is then transformed into a query vector using an embedding model for retrieval in a knowledge base. This knowledge base consists of relevant company documents and other reference documents. By associating fields such as "overtime hours" with sections like "overtime calculation rules" in the knowledge base, specific rule items such as "calculation multiplier" and "duration limit" can be extracted.

[0090] In some embodiments, the structured data to be audited, along with the retrieved and extracted rule items, can be distributed to multiple parallel, lightweight audit engines, each of which performs comparisons and calculations. Specifically, the attendance audit engine verifies whether the number of attendance days and overtime hours exceed the limits stipulated by regulations, such as a monthly overtime limit of 36 hours. The performance audit engine verifies whether performance levels match their corresponding coefficients and whether departmental performance distribution is compliant, such as the proportion of S-level employees not exceeding 10%. The salary compliance audit engine verifies whether actual wages are lower than the local minimum wage standard. The data consistency audit engine compares salary data with the employee roster in the company's human resources system, identifying anomalies such as "name mismatch," "department inconsistency," and "remaining paid salaries to departed employees."

[0091] In an optional embodiment, each specialized audit engine adopts a mode that combines rule-based judgment with a large language model. Based on rule-based judgment, the large language model is used to perform semantic understanding and analysis of fuzzy rules, thereby ensuring the accuracy and flexibility of the evaluation results.

[0092] In an optional embodiment, the evaluation results output by all audit engines can also be clustered and synthesized. For example, if the same employee has both "excessive overtime, medium risk" and "abnormal performance coefficient, medium risk" issues, the employee's overall risk level is upgraded from medium risk to high risk. Ultimately, data containing anomaly description information, risk level, department, and anomaly type is generated.

[0093] In an optional embodiment, an evaluation report for the target object is generated using a large language model, based on at least one evaluation result of the target object and a preset prompt word template. The preset prompt word template contains semantic elements of the evaluation report, including a summary, detailed information, classification statistics, improvement suggestions, and follow-up tracking plans. The evaluation report for the target object can be a directly submittable Markdown or HTML format report.

[0094] This embodiment also provides a data evaluation device, which can be integrated into a terminal device. For example, such as Figure 4 As shown, the data evaluation device may include: The acquisition module 401 is used to acquire the data to be evaluated associated with the target object and convert the data to be evaluated into structured data corresponding to the target object. The structured data includes at least one field and the field value corresponding to the field. The retrieval module 402 is used to search in a preset rule base based on the structured data corresponding to the target object to determine the set of rule items that match the target object; Evaluation module 403 is used to perform evaluation processing on at least one dimension based on structured data and a set of rule items, and to determine at least one evaluation result for the target object; Output module 404 is used to generate an evaluation report for the target object according to a preset evaluation report template based on at least one evaluation result of the target object.

[0095] The acquisition module 401, retrieval module 402, evaluation module 403 and output module 404 can be used to execute the corresponding embodiments of the above data evaluation method. For the specific implementation of these modules and more details, please refer to the corresponding method section, which will not be elaborated here.

[0096] In some embodiments of this application, the data evaluation apparatus can be implemented as a computer program, which can be implemented in, for example... Figure 5 The device operates on the electronic device shown. The memory of the electronic device can store various program modules that make up the data evaluation apparatus. The computer program composed of the various program modules causes the processor to execute the steps in the data evaluation methods of the various embodiments of this application described in the embodiments of this application.

[0097] Accordingly, this application also provides an electronic device, which can be a terminal, such as a smartphone, tablet computer, laptop computer, touch screen, game console, personal computer (PC), personal digital assistant (PDA), or other terminal device. Alternatively, the electronic device can be a server.

[0098] like Figure 5 As shown, Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device 500 includes a processor 501 with one or more processing cores, a memory 502 with one or more computer-readable storage media, and a computer program stored on the memory 502 and executable on the processor. The processor 501 and the memory 502 are electrically connected. Those skilled in the art will understand that the electronic device structure shown in the figure does not constitute a limitation on the electronic device, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0099] The processor 501 is the control center of the electronic device 500. It connects various parts of the electronic device 500 via various interfaces and lines. By running or loading software programs and / or units stored in the memory 502, and by calling data stored in the memory 502, it executes various functions and processes data of the electronic device 500, thereby providing overall monitoring of the electronic device 500. The processor 501 can be a central processing unit (CPU), a graphics processing unit (GPU), a network processor (NP), etc., and can implement or execute the methods, steps, and logic diagrams disclosed in the embodiments of this application.

[0100] In this embodiment of the application, the processor 501 in the electronic device 500 loads the instructions corresponding to the processes of one or more applications into the memory 502 according to the following steps, and the processor 501 runs the applications stored in the memory 502 to realize the various functions described in this embodiment of the application.

[0101] Optional, such as Figure 5 As shown, the electronic device 500 also includes: a touch display screen 503, a radio frequency circuit 504, an audio circuit 505, an input unit 506, and a power supply 507. The processor 501 is electrically connected to the touch display screen 503, the radio frequency circuit 504, the audio circuit 505, the input unit 506, and the power supply 507. Those skilled in the art will understand that... Figure 5 The electronic device structure shown does not constitute a limitation on the electronic device and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0102] The touch display screen 503 can be used to display a graphical user interface (GUI) and receive operation commands generated by the user interacting with the GUI. The touch display screen 503 may include a display panel and a touch panel. The display panel can be used to display information input by the user or information provided to the user, as well as various graphical user interfaces of the electronic device. These graphical user interfaces can be composed of graphics, text, icons, video, and any combination thereof. Optionally, the display panel can be configured using a liquid crystal display (LCD), organic light-emitting diode (OLED), or other similar technologies. The touch panel can be used to collect touch operations performed by the user on or near it (such as operations performed by the user using a finger, stylus, or any suitable object or accessory on or near the touch panel), generate corresponding operation commands, and execute the corresponding program according to the operation commands. Optionally, the touch panel may include two parts: a touch detection device and a touch controller. The touch detection device detects the user's touch location and the signal generated by the touch operation, transmitting the signal to the touch controller. The touch controller receives touch information from the touch detection device, converts it into touch point coordinates, and sends it to the processor 501. It can also receive and execute commands from the processor 501. The touch panel can cover the display panel. When the touch panel detects a touch operation on or near it, it transmits the information to the processor 501 to determine the dimension of the touch event. Subsequently, the processor 501 provides corresponding visual output on the display panel based on the dimension of the touch event. In this embodiment, the touch panel and the display panel can be integrated into the touch display screen 503 to achieve input and output functions. However, in some embodiments, the touch panel and the touch display screen 503 can be implemented as two independent components to achieve input and output functions. That is, the touch display screen 503 can also be used as part of the input unit 506 to achieve input functions.

[0103] The radio frequency circuit 504 can be used to transmit and receive radio frequency signals to establish wireless communication with network devices or other electronic devices, and to transmit and receive signals with network devices or other electronic devices.

[0104] Audio circuitry 505 can be used to provide an audio interface between a user and an electronic device via a speaker and a microphone. Audio circuitry 505 converts received audio data into electrical signals, transmits them to the speaker, and the speaker converts them into sound signals for output. Conversely, the microphone converts collected sound signals into electrical signals, which are then received by audio circuitry 505, converted back into audio data, and then processed by processor 501 before being transmitted via radio frequency circuitry 504 to, for example, another electronic device, or output to memory 502 for further processing. Audio circuitry 505 may also include an earphone jack to facilitate communication between peripheral headphones and electronic devices.

[0105] The input unit 506 can be used to receive input numbers, characters, or user characteristic information (such as fingerprints, iris, facial information, etc.), and to generate keyboard, mouse, joystick, optical, or trackball signal inputs related to user settings and function control.

[0106] Power supply 507 is used to supply power to various components of electronic device 500. Optionally, power supply 507 can be logically connected to processor 501 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. Power supply 507 may also include one or more DC or AC power supplies, recharging systems, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components.

[0107] although Figure 5 As not shown in the diagram, the electronic device 500 may also include a camera, sensor, wireless fidelity module, Bluetooth module, etc., which will not be described in detail here.

[0108] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0109] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0110] To this end, embodiments of this application provide a computer-readable storage medium storing a plurality of computer programs that can be loaded by a processor to execute any of the data evaluation methods provided in embodiments of this application. The computer program can execute the steps of the following data evaluation methods.

[0111] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0112] The computer-readable storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0113] Since the computer program stored in the computer-readable storage medium can execute any of the data evaluation methods provided in the embodiments of this application, the beneficial effects that any of the data evaluation methods provided in the embodiments of this application can achieve can be realized, as detailed in the preceding embodiments, and will not be repeated here.

[0114] According to one aspect of this application, a computer program product or computer program is also provided, comprising computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the methods provided in the various optional implementations of the above embodiments.

[0115] In the above embodiments of the data evaluation device, computer-readable storage medium, electronic device, and computer program product, the descriptions of each embodiment have different focuses. Parts not described in detail in a particular embodiment can be referred to in the relevant descriptions of other embodiments. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes and beneficial effects of the data evaluation device, computer-readable storage medium, computer program product, electronic device, and their corresponding units described above can be referred to the description of the data evaluation method in the above embodiments, and will not be repeated here.

[0116] The foregoing has provided a detailed description of a data evaluation method, apparatus, electronic device, computer-readable storage medium, and computer program product provided in the embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

[0117] It should be noted that, in the data processing stage, the technical solution of this application has strictly limited the scope of data collection to the minimum necessary to achieve the technical objectives, preventing the acquisition of irrelevant information. For any user information to be collected, the data subject will be clearly informed and their consent obtained. Furthermore, technologies such as encrypted storage and access control are employed to strengthen data security and ensure the security and compliance of the entire data processing process. The technical model and decision-making mechanism are based on objective technical parameters and do not introduce unnecessary parameters such as gender or age that may lead to discrimination, resolutely eliminating algorithmic discrimination and upholding public order and good morals. In addition, the specification fully describes the technical implementation methods, application scenarios, and compliance protection details. The claims are consistent with the content of the specification, key compliance designs are clear and verifiable, and the overall technical design is guided by the protection of public interests and adherence to social ethics, without any circumstances that harm public interests or violate public order and good morals.

Claims

1. A data evaluation method, characterized in that, The method includes: Obtain the data to be evaluated associated with the target object, and transform the data to be evaluated into structured data corresponding to the target object. The structured data includes at least one field and the field value corresponding to the field. Based on the structured data corresponding to the target object, a search is performed in a preset rule base to determine the set of rule items that match the target object; Based on the structured data and the set of rule items, perform evaluation processing on at least one dimension to determine at least one evaluation result for the target object; An evaluation report for the target object is generated according to a preset evaluation report template based on at least one evaluation result of the target object.

2. The data evaluation method according to claim 1, characterized in that, The step of performing evaluation processing based on the structured data and the set of rule items in at least one dimension to determine at least one evaluation result for the target object includes: Based on the dimension to which each evaluation process belongs, determine the set of sub-rule items required to complete each evaluation process from the set of rule items; Based on the set of sub-rule items required to complete each of the evaluation processes, the structured data is subjected to corresponding evaluation processes to determine the evaluation results corresponding to each of the evaluation processes.

3. The data evaluation method according to claim 2, characterized in that, The step of performing corresponding evaluation processing on the structured data based on the set of sub-rule items required to complete each of the evaluation processes, and determining the evaluation result corresponding to each of the evaluation processes, includes: Based on the set of sub-rule items corresponding to the evaluation process, the structured data is evaluated according to rules to obtain the first sub-evaluation result corresponding to the evaluation process; and / or, The structured data is semantically evaluated using a large language model based on the set of sub-rule items corresponding to the evaluation process, to obtain the second sub-evaluation result corresponding to the evaluation process. The evaluation result corresponding to the evaluation process is determined based on the first sub-evaluation result and / or the second sub-evaluation result.

4. The data evaluation method according to claim 3, characterized in that, The step of determining the evaluation result corresponding to the evaluation process based on the first sub-evaluation result and / or the second sub-evaluation result includes: Based on the distribution characteristics of the fuzzy rule items in the set of sub-rule items, determine the first weight corresponding to the first sub-evaluation result and the second weight corresponding to the second sub-evaluation result; The first sub-evaluation result and the second sub-evaluation result are weighted and fused according to the first weight and the second weight to obtain the evaluation result corresponding to the evaluation process.

5. The data evaluation method according to claim 1, characterized in that, The step of generating an evaluation report for the target object according to at least one evaluation result of the target object and a preset evaluation report template includes: The comprehensive evaluation result corresponding to the target object is determined by fusing the evaluation results corresponding to each evaluation process of the target object. Based on the comprehensive evaluation results corresponding to the target object, an evaluation report for the target object is generated according to a preset evaluation report template.

6. The data evaluation method according to claim 5, characterized in that, The comprehensive evaluation result includes the clustering anomaly information corresponding to the target object, and / or the comprehensive evaluation index corresponding to the target object. The step of fusing the evaluation results corresponding to each evaluation process of the target object to determine the comprehensive evaluation result for the target object includes: When the evaluation results contain corresponding anomaly information, semantic clustering is performed based on the anomaly information corresponding to each evaluation result to obtain the clustered anomaly information corresponding to the target object; and / or, Obtain the basic evaluation score and the influence coefficient corresponding to each evaluation result of the target object; Based on the basic evaluation score corresponding to each evaluation result and the influence coefficient corresponding to each evaluation result, the comprehensive evaluation index corresponding to the target object is obtained.

7. The data evaluation method according to any one of claims 1 to 6, characterized in that, The step of searching in a preset rule base based on the structured data corresponding to the target object to determine the set of rule items matching the target object includes: Based on the fields in the structured data and the preset mapping relationship, the business scenario tags corresponding to the fields are obtained, wherein the preset mapping relationship includes the preset business scenario tags corresponding to each preset field; Based on the business scenario tag, retrieve the rule description information associated with the business scenario tag from the preset rule base; The executable rule items are extracted from the rule description information using a large language model to obtain the rule item set.

8. A data evaluation device, characterized in that, The device includes: The acquisition module is used to acquire the data to be evaluated associated with the target object, and to convert the data to be evaluated into structured data corresponding to the target object. The structured data includes at least one field and the field value corresponding to the field. The retrieval module is used to search in a preset rule base based on the structured data corresponding to the target object to determine the set of rule items that match the target object; An evaluation module is used to perform evaluation processing on at least one dimension based on the structured data and the set of rule items, and to determine at least one evaluation result for the target object; The output module is used to generate an evaluation report for the target object according to a preset evaluation report template based on at least one evaluation result of the target object.

9. An electronic device, characterized in that, The system includes a processor and a memory, the memory storing multiple instructions; the processor loads instructions from the memory to perform the steps of the data evaluation method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a plurality of instructions adapted for loading by a processor to perform the steps of the data evaluation method as described in any one of claims 1 to 7.

Images