Enterprise performance data control system

Abstract

The invention relates to an enterprise performance data control system, and belongs to the technical field of performance data processing, and the system comprises a data collection module which is used for obtaining performance original data from a plurality of source systems in an enterprise, and carrying out the preprocessing of the original data; the index setting module is used for configuring a user-defined demand into a configuration file which can be analyzed by a computer in response to the user-defined demand; wherein the user-defined demand at least comprises a user-defined assessment scheme; the calculation engine module is used for acquiring the configuration file and executing performance assessment calculation according to the requirements of the configuration file to generate an assessment result; and the result display module is used for obtaining and visually displaying the assessment result, so that a user can obtain the assessment result. The method has the effects of data integration and flexible configuration of the assessment scheme.

Description

Technical Field

[0001] This application relates to the field of performance data processing technology, and in particular to an enterprise performance data control system. Background Technology

[0002] In modern corporate governance, performance appraisal is a core management tool for evaluating organizational and individual effectiveness and driving the achievement of strategic goals. Traditional performance management processes include performance planning, implementation, evaluation and feedback, and the application of performance data. Currently, many companies use ERP systems for performance management; however, these systems are typically used merely as document management tools, statically storing documented performance information. Furthermore, as companies grow and business complexity increases, traditional performance appraisal methods, due to their poor practicality and simplistic evaluation methods, can no longer meet the needs of modern enterprises and therefore require improvement. Summary of the Invention

[0003] In order to achieve the effects of data integration and flexible configuration of assessment schemes, this application provides an enterprise performance data control system.

[0004] Firstly, this application provides an enterprise performance data control system, comprising: The data acquisition module is used to obtain raw performance data from multiple source systems within the enterprise and to preprocess the raw data. The indicator setting module is used to respond to user-defined requirements and configure user-defined requirements into a configuration file that can be parsed by a computer; wherein, the user-defined requirements include at least user-defined assessment schemes; The calculation engine module is used to obtain the configuration file, perform performance evaluation calculations according to the requirements of the configuration file, and generate evaluation results; The results display module is used to acquire and visualize the assessment results so that users can know the assessment results.

[0005] By adopting the above technical solutions, raw performance data is obtained from various systems to achieve multi-source data integration, thereby optimizing the comprehensiveness and scientific nature of the assessment based on multi-source data analysis; the indicator setting module provides users with the function of customizing assessment standards, thereby realizing flexible configuration of assessment standards and improving the adaptability and flexibility of the system.

[0006] Optionally, the indicator setting module includes a demand interaction unit, an intent parsing unit, and a solution confirmation unit; The demand interaction unit is used to obtain user-defined demands, which may also include fuzzy intent information input by the user in natural language. The intent parsing unit is used to parse the user-defined requirements obtained by the requirement interaction unit, and generate and output a preliminary assessment plan based on the parsing results, so that the user can confirm the preliminary assessment plan. The scheme confirmation unit is used to configure the preliminary assessment scheme into a configuration file that can be parsed by a computer based on the user's confirmation opinion on the preliminary assessment scheme.

[0007] By adopting the above technical solution, users do not need to have professional knowledge of data modeling or indicator definition, nor do they need to understand the backend data structure. They can obtain professional assessment solutions simply through everyday expression. This expands the user group of the system from professional HR or data analysts to ordinary management, greatly improving the ease of use and popularity of the system. Moreover, compared with the cumbersome and time-consuming process of manually defining indicators and solutions in the traditional way, this solution can respond and generate complex solution suggestions within seconds, thereby improving customization efficiency.

[0008] Optionally, the intent parsing unit includes an assessment tendency inference subunit, a compliance check and correction subunit, and a fuzzy intent parsing subunit; The assessment tendency inference subunit is used to infer the user's assessment tendency reflected in the user-defined requirements; wherein, the assessment tendency includes at least the assessment object of the tendency and / or the assessment indicator of the tendency. The compliance check and correction subunit is used to perform a compliance check on the assessment scheme based on a pre-set business rule library when the received user-defined requirement is a user-defined assessment scheme. When a conflict is detected, modification suggestions are generated based on the user's evaluation preferences. Based on the modification suggestions, the user-defined assessment scheme is modified and an initial assessment scheme is generated and output. The fuzzy intent parsing subunit is used to parse the fuzzy intent information when the received user-defined requirement is fuzzy intent information input in natural language, and generate and output a preliminary assessment plan based on the parsing result and the user's evaluation tendency.

[0009] By adopting the above technical solutions, and by making "prejudgment of tendencies" and "conflict resolution" mandatory steps in all solution generation paths, the system ensures that the final output of the initial assessment solution not only conforms to the user's original intention but also meets the logical rigor and data feasibility of business rules, thereby improving the quality of the solution from the source. That is, the system no longer mechanically executes commands or simply recommends templates, but can proactively understand user preferences (assessment tendencies) and optimize solutions and resolve conflicts based on these preferences, making the output solution more in line with the user's real management scenarios and thinking patterns, thus achieving truly personalized decision support.

[0010] Optionally, it also includes an active learning and optimization module, which periodically reads the historically stored configuration files, processes the historically stored configuration files based on predefined organization and adjustment strategies, generates new candidate configuration files, and assigns a bias label to each candidate configuration file to characterize the assessment bias of the corresponding candidate configuration file. The active learning and optimization module is also used to use the assessment scheme corresponding to the candidate configuration file and the corresponding tendency label as training samples, and to use the training samples to conduct supervised learning training on the assessment tendency inference function of the assessment tendency inference subunit, so as to optimize the accuracy of the assessment tendency inference subunit inferring user assessment tendency based on assessment scheme.

[0011] By adopting the above technical solution, training data with accurate bias labels is automatically generated on a large scale on a regular basis based on historically stored configuration files. Then, the mapping relationship between "scheme-bias" is directly learned through supervised learning to optimize the inference accuracy of the bias inference subunit.

[0012] Optionally, it also includes a result correlation and attribution analysis module, which is used to periodically analyze the correlation between different assessment indicators based on raw data obtained from historical periods, and to construct an indicator relationship network. The result association and attribution analysis module is also used to identify abnormal assessment indicators after each assessment result is generated, and to search in the indicator relationship network whether there are any preceding indicators that are related to the abnormal assessment indicators. If so, the module analyzes the original data corresponding to the preceding indicators within the assessment period corresponding to the current assessment result. Based on the analysis results, the module adds an explanation of the abnormal data occurrence of the abnormal assessment indicators to the assessment results.

[0013] By adopting the above technical solution, instead of analyzing the abnormality of the corresponding assessment indicators based solely on the assessment data, we can analyze whether there are other factors that may cause the assessment indicators to be abnormal, such as the assessment indicators being affected by the corresponding preceding indicators. This allows for a more objective and comprehensive assessment, avoiding erroneous decisions based on superficial data.

[0014] Optionally, it also includes a process monitoring and attribution analysis module, which is used to periodically extract phased assessment indicator data during the assessment cycle to form time series data of assessment indicators, and to perform phased analysis on the time series data, wherein the phased analysis includes at least data trend analysis. The process monitoring and attribution analysis module is also used to generate and send intervention reminder information before the end of the assessment period when the phased analysis results are abnormal, so that users can be informed.

[0015] By adopting the above technical solution, the assessment indicator data can be monitored in real time during the assessment period. When the assessment indicator data is abnormal, an early warning will be issued to remind users to intervene in time. This may reverse the unfavorable situation before the end of the assessment period and reduce the probability of poor assessment results.

[0016] Optionally, the result association and attribution analysis module is also used to send linkage information to the process monitoring and attribution analysis module when an abnormal assessment indicator is identified, so that the process monitoring and attribution analysis module can confirm whether the process monitoring and attribution analysis module has generated intervention reminder information corresponding to the abnormal assessment indicator within the assessment period corresponding to the abnormal assessment indicator. If so, the process monitoring and attribution analysis module sends the phased analysis results of generating the corresponding intervention reminder information to the result association and attribution analysis module. The result association and attribution analysis module is also used to, upon receiving interim analysis results, use the interim analysis results as explanations for the abnormal assessment indicators and add them to the corresponding assessment results.

[0017] By adopting the above technical solution, if abnormal performance indicators have experienced abnormal situations within the assessment period, the corresponding interim analysis results will also be used as explanations. This makes the assessment results provided to users no longer just a simple score and a vague explanation, but a "diagnostic report" with a timeline. This helps users to more comprehensively trace the specific changes in abnormal performance indicators and provides a very clear direction for subsequent improvements.

[0018] Secondly, this application provides a method for controlling enterprise performance data. Raw performance data is obtained from multiple source systems within the enterprise, and the raw data is preprocessed. In response to user-defined requirements, the user-defined requirements are configured into a configuration file that can be parsed by a computer; wherein, the user-defined requirements include at least a user-defined assessment scheme; Obtain the configuration file and perform performance evaluation calculations according to the requirements of the configuration file to generate evaluation results; The assessment results are acquired and visualized so that users can be informed of them.

[0019] Thirdly, this application provides an enterprise performance data control device, including a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described in the second aspect.

[0020] Fourthly, this application provides a computer-readable storage medium storing a computer program that can be loaded by a processor and execute the method described in the second aspect.

[0021] In summary, this application includes the following beneficial technical effects: In this application, raw performance data is obtained from various systems to achieve multi-source data integration, thereby optimizing the comprehensiveness and scientific nature of the assessment based on multi-source data assessment analysis; the indicator setting module provides users with the function of customizing assessment standards, thereby realizing flexible configuration of assessment standards and improving the adaptability and flexibility of the system. Furthermore, users do not need professional knowledge of data modeling or indicator definition, nor do they need to understand the backend data structure. They can obtain professional assessment solutions simply through everyday expression. This expands the user base of the system from professional HR or data analysts to ordinary management, greatly improving the system's ease of use and accessibility. Compared to the cumbersome and time-consuming process of manually defining indicators and solutions in the traditional way, this solution can respond and generate complex solution suggestions within seconds, thereby improving customization efficiency. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention, 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 the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a structural block diagram of the enterprise performance data control system disclosed in the embodiments of this application.

[0024] Figure labeling: 1. Data acquisition module; 2. Indicator setting module; 21. Requirement interaction unit; 22. Intent parsing unit; 221. Assessment tendency inference sub-unit; 222. Compliance check and correction sub-unit; 223. Fuzzy intent parsing sub-unit; 23. Solution confirmation unit; 3. Calculation engine module; 4. Result display module; 5. Active learning and optimization module; 6. Result correlation and attribution analysis module; 7. Process monitoring and attribution analysis module. Detailed Implementation

[0025] The following is in conjunction with the appendix Figure 1 This application will be described in further detail.

[0026] This application discloses an enterprise performance data control system. (Refer to...) Figure 1 It includes a data acquisition module 1, an indicator setting module 2, a calculation engine module 3, and a result display module 4. Among them: Data acquisition module 1 is used to acquire raw performance data from multiple source systems within the enterprise and to preprocess the raw data; The indicator setting module 2 is used to respond to user-defined needs and configure user-defined needs into a configuration file that can be parsed by a computer; wherein, user-defined needs include at least user-defined assessment schemes; The calculation engine module 3 is used to obtain the configuration file and perform performance evaluation calculations according to the requirements of the configuration file to generate the evaluation results; Result Display Module 4 is used to acquire and visualize the assessment results so that users can know the assessment results.

[0027] In implementation, data acquisition module 1 is configured to establish communication connections with existing source systems within the enterprise (such as financial systems and human resources systems) through predefined API interfaces (e.g., RESTful API, SOAP API, etc.). Data acquisition module 1 retrieves raw performance-related data from each source system in real-time or near real-time according to predetermined time periods (e.g., every minute) or based on event triggers (e.g., push notifications when source system data is updated). The corresponding raw data includes, but is not limited to: sales revenue, costs, profits, attendance rates, project completion progress, etc.

[0028] The data acquisition module 1 also includes a data cleaning and standardization submodule, which cleans the acquired heterogeneous data according to predefined rules (removing duplicates and handling null values), converts the format (such as unifying the date format to YYYY-MM-DD), and unifies the units (such as unifying the amount to RMB "yuan"), and finally generates standardized data records that conform to the target pattern and writes them to the preset central database. The central database 102 preferably adopts a relational database (such as MySQL or PostgreSQL) or a large-scale distributed database (such as Hadoop HBase).

[0029] The Indicator Setting Module 2 provides a graphical user interface (GUI) for users (such as HR or managers) to customize performance indicators and assessment schemes without writing code. Users can set indicator names (e.g., "Department Quarterly Sales Achievement Rate") and indicator formulas (e.g., "Actual Sales / Target Sales * 100%)" by dragging and dropping components or filling out forms. Variables in the indicator formulas are mapped to specific data fields in the central database. Multiple performance indicators are then combined into an assessment scheme, with weights assigned to each indicator, assessment periods (monthly, quarterly, annual), and target groups (departments, individuals). After user completion, the Indicator Setting Module 2 serializes the configuration information (formulas, weights, periods, etc.) into a computer-parseable configuration file (e.g., JSON, XML, or YAML format) and stores it in the central database. Furthermore, users can select any configuration file from the central database as the primary configuration file, and its corresponding assessment scheme as the primary assessment scheme.

[0030] The calculation engine module 3 is used to obtain the main configuration file and parse the configuration information (formulas, weights, periods, etc.) defined in it. Then, according to the period specified in the configuration information, it extracts the raw data of the corresponding period and range from the central database, calls the built-in formula parser and calculation component, executes the formula in the configuration file, and performs weighted summation in combination with the weights, and finally generates the performance evaluation result data (i.e., evaluation results) for each evaluation object.

[0031] The Results Display Module 4 is used to obtain the assessment results generated by the Computation Engine Module 3 and provide users with a visual display through a web interface or client application. Display formats include, but are not limited to, bar charts, line charts, pie charts, radar charts, and data tables. The Results Display Module 4 supports multi-dimensional dynamic analysis. Users can independently select different analysis dimensions (such as time, department, and job level) through interactive controls such as drop-down menus and filters on the interface. The Results Display Module 4 then dynamically generates and refreshes charts, enabling drill-down, roll-up, slicing, and block analysis of the data, thereby providing in-depth decision-making insights.

[0032] Optionally, this application proposes that user-defined requirements can be assessment schemes with specific performance indicator names and formulas provided by professionals, or fuzzy intent information input by users in natural language, such as "Compare the performance of the Beijing and Shanghai teams in the third quarter" or "Find the common characteristics of salespeople with high customer satisfaction." To this end, this application will provide corresponding processing operations for user-defined requirements of the aforementioned two different input types. For example, for assessment schemes with specific performance indicator names and formulas provided by professionals, this application proposes to conduct compliance checks to determine whether there are any obviously illogical issues or conflicts between indicators (such as the total weight percentage in a weighted summation not equaling 100%), thereby correcting the assessment scheme. For fuzzy intent information input in natural language, this application will parse it and automatically generate corresponding assessment schemes for user verification.

[0033] Accordingly, in order to implement the above solution, the indicator setting module 2 of this application specifically includes a demand interaction unit 21, an intent parsing unit 22, and a solution confirmation unit 23.

[0034] Specifically, the requirement interaction unit 21 provides a graphical user interface (GUI) for users (such as HR or managers) to input user-defined requirements (i.e., obtain user-defined requirements).

[0035] The intent parsing unit 22 is used to parse the user-defined requirements and obtain a preliminary assessment plan, which is then passed to the plan confirmation unit 23 to output and display the preliminary assessment plan. This further enables interaction, allowing the user to input confirmation opinions (such as "confirmed without error" or to further modify certain indicators or plans based on the preliminary assessment plan). Next, the plan confirmation unit 23 uses the obtained confirmation opinions. If the confirmation opinion is "confirmed without error", the corresponding preliminary assessment plan is used as the final assessment plan. If the confirmation opinion includes content that the user has further modified, the preliminary assessment plan is readjusted according to the user's modifications to obtain the final assessment plan. Finally, it is configured into a configuration file according to the method disclosed above.

[0036] The intent parsing unit 22 further includes an assessment tendency inference subunit 221, a compliance check and correction subunit 222, and a fuzzy intent parsing subunit 223.

[0037] The assessment tendency inference subunit 221 is used to infer the user's assessment tendency reflected in the user-defined requirements; wherein, the assessment tendency includes at least the assessment object of the tendency and the assessment indicator of the tendency.

[0038] In implementation, the performance evaluation bias inference subunit 221 pre-constructs an enterprise organizational structure knowledge graph, including hierarchical relationships such as departments, teams, and job levels. The performance evaluation bias inference subunit 221 is used to infer the preferred performance evaluation targets based on the aforementioned enterprise organizational structure knowledge graph. Specifically, the performance evaluation bias inference subunit 221 scans the user's custom requirements (raw natural language text or parsed and generated performance evaluation schemes with specific performance indicator names and formulas) and identifies whether the user's custom requirements repeatedly mention or explicitly specify a certain organizational entity name from the enterprise organizational structure knowledge graph (such as "sales department" or "post-90s employees"). If so, it is directly identified as the preferred performance evaluation target; if not, that is, the content of the user's custom requirements is relatively general (such as "analyze efficiency," which does not contain an organizational entity name), then the user's historical operation records (such as which departments' data they frequently view) or job level (such as the sales director's default preferred target being the sales system) are analyzed to determine the preferred performance evaluation target.

[0039] Furthermore, the performance evaluation tendency inference subunit 221 also has a pre-defined indicator type library. This library stores all predefined performance indicator names and their category tags (such as "financial", "operations", and "customer" categories, with each category tag also having its own knowledge graph for storing different descriptive terms used to describe that category tag; for example, "financial" can also be described as "cost"). It also includes the performance evaluation objects corresponding to each category tag; for example, the performance evaluation objects corresponding to "financial" include "sales department". Additionally, for performance indicators with the same category tag, the correlation between indicators can be pre-analyzed manually. Indicators that reflect the same business concepts (such as "efficiency", "quality", and "cost") are identified as highly correlated indicators, and their correlation relationships are stored.

[0040] The performance evaluation preference inference subunit 221 is also used to infer the performance evaluation indicators based on the indicator type library. Specifically, the performance evaluation preference inference subunit 221 scans the user's custom requirements (raw natural language text or parsed and generated performance evaluation schemes with specific performance indicator names and formulas), and identifies whether there are business concepts (such as "efficiency," "quality," and "cost") in the user's custom requirements. It then performs semantic matching between the identified business concepts and the tags and performance indicator names in the indicator type library. For example, if the user mentions "efficiency," it may match indicators such as "output per capita" and "project cycle completion rate." Finally, the system outputs a list of indicators that the user may prefer, sorted by matching degree, which serves as the performance evaluation indicator of preference.

[0041] Optionally, the compliance check and correction subunit 222 parses and processes user-defined requirements (provided by professionals) containing specific assessment indicator names and formulas, generating a preliminary assessment plan. Meanwhile, the fuzzy intent parsing subunit 223 parses and processes user-defined requirements (inputted in natural language) containing fuzzy intent information, also generating a preliminary assessment plan. Therefore, when the requirement interaction unit 21 receives a user-defined requirement, it can determine the type of the requirement. If it only contains assessment indicators and formulas, the requirement is sent to the compliance check and correction subunit 222 for processing; otherwise, it is sent to the fuzzy intent parsing subunit 223 for processing.

[0042] Specifically, the compliance check and correction subunit 222 is used to perform a compliance check on the assessment scheme based on a pre-set business rule library when the received user-defined requirement is a user-defined assessment scheme. When a conflict is detected, modification suggestions are generated based on the user's evaluation preferences. Based on the modification suggestions, the user-defined assessment scheme is modified and an initial assessment scheme is generated and output.

[0043] In implementation, the compliance check and correction subunit 222 has a pre-built business rule base. For example, the rules stored in the business rule base can exist in the form of "IF-THEN-ACTION," and each rule can be formally represented as a triple (trigger condition, conflict type, priority). The trigger condition describes a business or technical state that will cause a conflict; it is a logical expression based on scheme elements (such as assessment period / assessment indicator / weight value) and system metadata (such as the data update frequency of assessment indicators). The conflict type defines the nature of the conflict (such as "sequential conflict" or "logical conflict"), and the priority defines the severity of the conflict (such as "error" or "warning"). For example, the corresponding rule content can be as follows: IF (The assessment period in the adaptive requirements is "monthly", and the data update frequency of indicator A in the assessment plan is "quarterly"), THEN priority = "ERROR", conflict type = "The data of indicator A cannot support the monthly assessment"; IF (The sum of the weights of all performance indicators in the adaptive requirements ≠ 100%) THEN Priority = "ERROR"; IF (Historical data coverage of performance indicator B in adaptive requirements < 80%) THEN Priority = "WARNING", Conflict Type = "Incomplete data for indicator B, calculation results may be distorted".

[0044] The compliance check and correction subunit 222 is used to break down user-defined requirements into the smallest elements (assessment indicators, cycles, weights, target values) and match them one by one with the rules in the business rule base. If a rule is matched (i.e. the corresponding rule is satisfied), a conflict example is generated, and the conflict type and triggering conditions in the corresponding rule are recorded.

[0045] At this point, the assessment tendency determined by the assessment tendency inference subunit 221 is invoked to generate personalized modification suggestions. The user-defined requirements are then automatically modified according to these suggestions to obtain a preliminary assessment plan. It should be noted that this generation process prioritizes ensuring that the modified plan (i.e., the preliminary assessment plan) maintains, to the greatest extent possible, the core intent reflected in the user's original customized requirements.

[0046] Specifically, the compliance check and correction sub-unit 222 addresses conflicting indicators (e.g., in the original plan, the user's highly concerned indicator is "project gross profit margin," but a conflict arises due to complex data calculation and update delays; the corresponding component is "project gross profit margin"). It searches the indicator type library for alternative indicators with similar functions and semantics that align with the user's performance evaluation preferences (e.g., preferred indicator types and evaluation targets). For example, if the user's preferred performance indicators are profitability and project dimensions, it finds highly relevant indicators such as "project collection rate" or "cost budget achievement rate," which also reflect efficiency but have more readily available data (i.e., faster data update frequency), as alternative components. These alternative indicators replace the conflicting indicators in the user's custom requirements, thereby modifying the plan and obtaining the initial evaluation scheme. Alternatively, it can directly adjust the content of the user's custom requirements to resolve conflicts, such as adjusting the evaluation cycle: based on the data update frequency of all evaluation indicators in the custom requirements, a new cycle that covers all data update frequencies is calculated.

[0047] The following is an example of conflict detection and scheme modification in the compliance check and correction sub-unit 222: When detecting conflicts, if a rule containing the content "Data for indicator A cannot support monthly assessment" is matched, a data update frequency conflict is detected. In this case, a modification suggestion is generated based on the assessment preference (e.g., the preferred assessment target is the R&D department, and the preferred assessment indicator is "project delivery quality"): "You set the assessment period to 'monthly', but the 'project delivery quality' data comes from customer acceptance and the update period is 'after project completion'. Based on your preference for focusing on the R&D team's project results, it is recommended to adjust the assessment period to 'quarterly' and use it in conjunction with 'monthly code commits' (a phased indicator)."

[0048] The fuzzy intent parsing subunit 223 is used to parse the fuzzy intent information when the received user-defined requirement is fuzzy intent information input in natural language, and generate and output the preliminary assessment plan based on the parsing result and the user's evaluation tendency.

[0049] In implementation, the fuzzy intent parsing subunit 223 incorporates a natural language processing (NLP) engine (based on pre-trained models such as BERT) to perform word segmentation, semantic understanding, and entity recognition on the text input by the user in natural language (i.e., fuzzy intent information, such as "compare the performance of the Beijing and Shanghai teams in the third quarter" mentioned above). It then extracts key elements, including: analysis objectives (such as "return on investment," "common features," "predict...achieving performance targets"), analysis dimensions (such as "team," "salesperson," "product line"), filtering conditions (such as "Shanghai," "last quarter," "high customer satisfaction"), and operation types (such as "analyze," "find," "predict"). Furthermore, the fuzzy intent parsing subunit 223 is internally connected to an indicator knowledge base, which stores: a predefined library of standard indicators (such as "return on investment," "sales revenue," "customer satisfaction score"), relationships between indicators (such as a positive correlation between "customer satisfaction" and "sales revenue"), and commonly used performance evaluation templates.

[0050] The fuzzy intent parsing subunit 223 is used to perform matching and logical reasoning in the indicator knowledge base based on the parsed key elements, generating one or more concrete assessment scheme suggestions. For example, "Compare the performance of the Beijing and Shanghai teams in the third quarter" corresponds to the operation type "comparison", the analysis dimension "team", the conditions "Beijing, Shanghai, third quarter", and the analysis target "performance". Then, using the operation type, analysis dimension, conditions, and analysis target as search conditions, it queries and matches in the indicator knowledge base to obtain a preliminary set of candidate indicators with a relatively wide range. For example, it matches indicators related to "performance" such as "sales revenue", "sales revenue achievement rate", and "number of new customers" from the indicator knowledge base.

[0051] Next, the assessment tendency inferred by the assessment tendency inference subunit 221 is invoked to eliminate assessment indicators in the candidate indicator set that are obviously irrelevant to the user's assessment tendency. For example, if the assessment target of the tendency is "front-end department", then indicators in the candidate indicator set that are pre-marked as only applicable to "production department" are eliminated. Then, the relevance score of each indicator in the candidate indicator set to the user's assessment tendency is calculated. The score is based on three dimensions: semantic similarity (the textual similarity between the name, definition and description of the indicator and keywords in the assessment tendency (such as "efficiency" and "quality")), historical co-occurrence frequency (the frequency with which the indicator has been used in the same scheme as the same assessment tendency in history), and object applicability (whether the indicator can be applied to the assessment target of the determined tendency), as well as the pre-set weight values ​​of these three dimensions. The relevance score is finally obtained by weighted summation.

[0052] The indicators in the candidate indicator set are sorted in descending order of relevance score, and the top-ranked indicators are selected as the core indicator set. An initial weight is then assigned to each indicator (an equal weighting method can be used). An assessment period is set (prioritizing the period mentioned in the semantic analysis results; if not mentioned, the data update frequency of the core indicator set is analyzed to select a period that covers all indicator data and conforms to the regular management rhythm of the assessed entity). Finally, data source mapping is performed (automatically associating indicators with specific data table fields in the central database). All configured information is then integrated and assembled into a complete and executable preliminary assessment plan.

[0053] Optionally, it also includes an active learning and optimization module 5, which periodically reads the historically stored configuration files, processes the historically stored configuration files based on predefined organization and adjustment strategies, generates new candidate configuration files, and assigns a tendency label to each candidate configuration file to characterize the assessment tendency of the corresponding candidate configuration file. The active learning and optimization module 5 is also used to use the assessment scheme corresponding to the candidate configuration file and the corresponding tendency label as training samples, and to use the training samples to supervise the assessment tendency inference function of the assessment tendency inference subunit 221 to optimize the accuracy of the assessment tendency inference subunit 221 in inferring the user's assessment tendency based on the assessment scheme.

[0054] In implementation, the proactive learning and optimization module 5 periodically reads configuration files stored in the central database, employing a conscious combination and adjustment strategy, rather than complete randomness, to ensure that newly generated configuration files have clear bias labels and establish a correspondence between newly generated configuration files and bias labels (e.g., {biased assessment target: "Sales Department", biased assessment indicator: "Finance"}). The corresponding combination and adjustment strategies include: a. Commonality-based combinations: Identify frequently co-occurring groups of performance indicators (e.g., "sales revenue," "gross profit margin," and "collection rate" often appear together) or performance periods (e.g., "quarterly" performance evaluations are often linked to "project-based" indicators) in historical configuration files. Cross-merge configuration files containing the same commonality combinations (e.g., performance indicator groups, performance periods) to generate new schemes. For example, apply the weight allocation method of scheme A (containing indicator group X) to the indicator set of scheme B (containing the same indicator group X).

[0055] b. Scenario-based Adjustment: Several specific scenarios are predefined (scenarios are used to characterize assessment preferences, such as "cost control scenario," "market expansion scenario," and "R&D innovation scenario"). Before adjusting the indicators or parameters of the entire configuration file, the target scenario is clearly specified, and then the indicators or parameters of the corresponding configuration file are adjusted according to the target scenario. For example, for an assessment scheme corresponding to an existing configuration file, all its weights are tilted towards "cost-related" indicators (i.e., "cost-related" assessment indicators are given higher weights), thereby generating a new scheme clearly labeled as "cost control preference," where the corresponding preference assessment indicator is "financial." It should be noted that the preference label determined here may only include the preference assessment object, the preference assessment indicator, or both. In other embodiments, the preference assessment object can be further derived based on the determined preference assessment indicator. For example, the assessment object related to "financial" here includes "sales department."

[0056] Next, all the generated configuration files and their corresponding bias labels are combined into a training sample set. Each sample can be in the format: (assessment scheme corresponding to the configuration file, bias label). Then, using the assessment schemes in the training sample set as input and the bias labels as output, supervised learning training is performed on the machine learning model of the assessment bias inference subunit 221. The training process is as follows: The assessment schemes in the sample are input into the machine learning model of the assessment tendency inference subunit 221. The machine learning model of the assessment tendency inference subunit 221 outputs an assessment tendency. The difference between the assessment tendency and the tendency label stored in the sample is calculated. The model parameters are adjusted through the backpropagation algorithm to continuously reduce this difference.

[0057] Optionally, it also includes a result association and attribution analysis module 6, which is used to periodically analyze the correlation between different assessment indicators based on the raw data obtained in historical periods, and to construct an indicator relationship network. The results association and attribution analysis module 6 is also used to identify abnormal assessment indicators after each assessment result is generated. In the indicator relationship network, it searches for whether there are any preceding indicators that are related to the abnormal assessment indicators. If so, it analyzes the original data corresponding to the preceding indicators within the assessment period corresponding to the current assessment result. Based on the analysis results, it adds an explanation of the abnormal data of the abnormal assessment indicators to the assessment results.

[0058] In implementation, the Result Correlation and Attribution Analysis Module 6 is used to analyze the statistical relationships between all performance indicators based on the raw data collected from various source systems during historical periods. Analysis methods include: correlation analysis (calculating Pearson correlation coefficient, Spearman rank correlation coefficient, etc., to identify linear and nonlinear relationships), causal discovery algorithms (using methods such as PC algorithm and Granger causality test to attempt to infer the direction of potential causal relationships between performance indicators), and business rule injection (injecting known business rules (such as "sales revenue" directly affecting "profit") as prior knowledge into the control system). The final analysis output is a network diagram of indicator relationships, where nodes are performance indicators, and edges represent the strength and positive / negative correlation between indicators. For example, there is a strong negative correlation between "customer complaint volume" and "customer satisfaction," and a positive correlation between "market investment expenses" and "new customer numbers."

[0059] Whenever the calculation engine module 3 generates an assessment result, and before the result display module 4 displays the assessment result, the result association and attribution analysis module 6 is used to analyze whether there are any indicators in the assessment result that are significantly lower or higher than the expected assessment indicators (i.e., abnormal assessment indicators). Here, the expected range can be the specific expected range of each assessment indicator entered by the user when inputting custom requirements, or a data range that the control system can generate autonomously based on the maximum and minimum values ​​of the original data of the same assessment indicator and the same assessment object in historical periods as the range endpoint values.

[0060] For abnormal performance indicators, the Result Association and Attribution Analysis Module 6 searches the indicator relationship network for a strong correlation between the abnormal performance indicator and the preceding indicator. The preceding indicator refers to the performance indicator that influences the abnormal performance indicator. If such a correlation exists, the Result Association and Attribution Analysis Module 6 extracts the original data of the preceding indicator within the specified performance period of the performance plan corresponding to the current performance result. It analyzes the original data of the preceding indicator (i.e., checks the label of the preceding indicator within the performance period). If the preceding indicator data also shows a significant drop below or above expectations, indicating an abnormal preceding indicator, the explanation "The preceding indicator is abnormal, thus causing the corresponding abnormal performance indicator data to be abnormal" is added to the performance result. For example, if "Team A's 'Customer Satisfaction' score drops significantly," the Result Association and Attribution Analysis Module 6 finds a high negative correlation between "Customer Satisfaction" and "Customer Response Time" in the indicator relationship network and checks that the "Customer Response Time" indicator significantly increases within the corresponding performance period. Therefore, it can generate the explanation: "Team A's 'Customer Satisfaction' drop is mainly likely due to the increased 'Customer Response Time' within this performance period."

[0061] Finally, after the result association and attribution analysis module 6 completes the above analysis and processing of the assessment results, the processed assessment results are sent to the result display module 4 so that the result display module 4 can display the corresponding assessment results.

[0062] Optionally, it also includes a process monitoring and attribution analysis module 7, which is used to periodically extract phased assessment indicator data during the assessment cycle, form time series data of assessment indicators, and perform phased analysis on the time series data. The phased analysis includes at least data trend analysis. The process monitoring and attribution analysis module 7 is also used to generate and send intervention reminder information before the end of the assessment period when the phase analysis results are abnormal, so that users can be informed.

[0063] The result association and attribution analysis module 6 is also used to send linkage information to the process monitoring and attribution analysis module 7 when abnormal assessment indicators are identified, so that the process monitoring and attribution analysis module 7 can confirm whether the process monitoring and attribution analysis module 7 has generated intervention reminder information corresponding to the abnormal assessment indicator within the assessment period corresponding to the abnormal assessment indicator. If so, the process monitoring and attribution analysis module 7 will send the phased analysis results of generating the corresponding intervention reminder information to the result association and attribution analysis module 6. The results association and attribution analysis module 6 is also used to, when receiving interim analysis results, use the interim analysis results as explanations for the abnormal assessment indicators and add them to the corresponding assessment results.

[0064] In implementation, the raw data collected by data acquisition module 1 from each source system is acquired in real time. Whenever indicator setting module 2 generates a configuration file, it sends the file to both the calculation engine module 3 and the process monitoring and attribution analysis module 7. The calculation engine module 3 retrieves the data for the corresponding assessment indicators within the assessment period at the end of the specified assessment period and calculates the assessment results. Meanwhile, the process monitoring and attribution analysis module 7 acquires the data for each assessment indicator collected in real time by data acquisition module 1 during the assessment period, generates a time series data set for each indicator, and performs the following phased analyses on each time series data set: a. Trend judgment: Use methods such as linear regression or moving average to determine whether the performance indicators are in an upward trend, a downward trend, or a stable fluctuation. b. Outlier detection: Using algorithms such as Statistical Process Control (SPC) or Isolation Forest, outlier data points that significantly deviate from the normal fluctuation range are identified; c. Predictive Analysis: Based on the trend of the corresponding assessment data before the current time, use a time series forecasting model to predict the final value of the corresponding assessment indicator before the end of the assessment period.

[0065] The process monitoring and attribution analysis module 7 is used to determine in real time whether the phased analysis results meet the preset early warning rules based on the aforementioned real-time phased analysis results. If they do, the phased analysis results are considered abnormal, and an intervention reminder message is generated and sent to the user terminal so that the user can be informed and make timely adjustments. Here, the user terminal can refer to the smart terminal used by the user who proposed the user-defined requirements of the assessment scheme corresponding to the current configuration file. An example of the corresponding early warning rules is as follows: If a performance indicator deviates from the expected trend, or if more than a preset number of outliers occur within a specified time period, or if the final value of the predicted analysis is less than the target value * X% (where X is a preset positive number), an intervention alert will be generated. The expected trend mentioned here refers to the pre-set trend for the performance indicator. For example, the preset trend for "customer satisfaction" is an upward trend, while the preset trend for "customer complaints" is a downward trend.

[0066] In addition, the result association and attribution analysis module 6 is also used to send linkage information to the process monitoring and attribution analysis module 7 when abnormal assessment indicators are detected, so that the process monitoring and attribution analysis module 7 can confirm whether intervention reminder information has been generated for abnormal assessment indicators during the assessment period. If so, the process monitoring and attribution analysis module 7 will send the phased analysis results of the corresponding abnormal assessment indicator when the intervention reminder information was generated to the result association and attribution analysis module 6. The result association and attribution analysis module 6 is used to add the phased analysis results as an explanation to the assessment results when it receives the phased analysis results corresponding to the abnormal assessment indicators.

[0067] This application also discloses a method for controlling enterprise performance data, including the following steps: Raw performance data is obtained from multiple source systems within the enterprise, and the raw data is preprocessed. In response to user-defined requirements, the user-defined requirements are configured into a configuration file that can be parsed by a computer; wherein, the user-defined requirements include at least the user-defined assessment scheme; Obtain the configuration file and, according to the requirements of the configuration file, perform performance evaluation calculations and generate evaluation results; Acquire and visualize the assessment results so that users can be informed of them.

[0068] This application also discloses an enterprise performance data control device, which includes a memory and a processor. The memory stores a computer program that can be loaded by the processor and executed as described above for enterprise performance data control.

[0069] This application also discloses a computer-readable storage medium that stores a computer program that can be loaded by a processor and executed as described above in the enterprise performance data control method. The computer-readable storage medium includes, for example, various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0070] It should be noted that in this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0071] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit the scope of protection of the application. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on these embodiments, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

Claims

1. A corporate performance data control system, characterized in that, include: The data acquisition module (1) is used to acquire raw performance data from multiple source systems within the enterprise and to preprocess the raw data. The indicator setting module (2) is used to respond to user-defined requirements and configure user-defined requirements into a configuration file that can be parsed by a computer; wherein, the user-defined requirements include at least user-defined assessment schemes; The calculation engine module (3) is used to obtain the configuration file and perform performance evaluation calculations according to the requirements of the configuration file to generate evaluation results; The results display module (4) is used to obtain and visualize the assessment results so that users can know the assessment results.

2. The enterprise performance data control system according to claim 1, characterized in that, The indicator setting module (2) includes a demand interaction unit (21), an intent parsing unit (22), and a solution confirmation unit (23); The demand interaction unit (21) is used to obtain user-defined demands, which may also include fuzzy intent information input by the user in natural language. The intent parsing unit (22) is used to parse the user-defined requirements obtained by the requirement interaction unit (21), and generate and output a preliminary assessment plan based on the parsing results, so that the user can confirm the preliminary assessment plan. The scheme confirmation unit (23) is used to configure the preliminary assessment scheme into a configuration file that can be parsed by a computer based on the user's confirmation opinion on the preliminary assessment scheme when receiving the user's confirmation opinion on the preliminary assessment scheme.

3. The enterprise performance data control system according to claim 2, characterized in that, The intent parsing unit (22) includes an assessment tendency inference subunit (221), a compliance check and correction subunit (222), and a fuzzy intent parsing subunit (223); The assessment tendency inference subunit (221) is used to infer the user's assessment tendency reflected in the user-defined requirements; wherein, the assessment tendency includes at least the assessment object of the tendency and / or the assessment indicator of the tendency. The compliance check and correction subunit (222) is used to check the compliance of the assessment scheme based on the preset business rule library when the received user-defined requirement is a user-defined assessment scheme. When a conflict is detected, it generates modification suggestions based on the user's evaluation tendency, and modifies the user-defined assessment scheme based on the modification suggestions to generate and output the initial assessment scheme. The fuzzy intent parsing subunit (223) is used to parse the fuzzy intent information when the received user-defined requirement is fuzzy intent information input in natural language, and generate and output a preliminary assessment plan based on the parsing result and the user's evaluation tendency.

4. The enterprise performance data control system according to claim 3, characterized in that, It also includes an active learning and optimization module (5), which periodically reads the historically stored configuration files, processes the historically stored configuration files based on predefined organization and adjustment strategies, generates new candidate configuration files, and assigns a tendency label to each candidate configuration file to characterize the assessment tendency of the corresponding candidate configuration file. The active learning and optimization module (5) is also used to use the assessment scheme corresponding to the candidate configuration file and the corresponding tendency label as training samples, and to use the training samples to supervise the assessment tendency inference function of the assessment tendency inference subunit (221) to optimize the accuracy of the assessment tendency inference subunit (221) in inferring the user's assessment tendency based on the assessment scheme.

5. The enterprise performance data control system according to claim 1, characterized in that, It also includes a result association and attribution analysis module (6), which is used to periodically analyze the correlation between different assessment indicators based on the raw data obtained in historical periods, and to construct an indicator relationship network; The result association and attribution analysis module (6) is also used to identify abnormal assessment indicators after each assessment result is generated, and to search in the indicator relationship network whether there are any preceding indicators that are related to the abnormal assessment indicators. If there are, the original data corresponding to the preceding indicators in the assessment period corresponding to the current assessment result are analyzed. Based on the analysis results, an explanation of the abnormal data for the abnormal assessment indicators is added to the assessment results.

6. The enterprise performance data control system according to claim 5, characterized in that, It also includes a process monitoring and attribution analysis module (7), which is used to periodically extract phased assessment indicator data during the assessment cycle, form time series data of assessment indicators, and perform phased analysis on the time series data. The phased analysis includes at least data trend analysis. The process monitoring and attribution analysis module (7) is also used to generate and send intervention reminder information before the end of the assessment period when the phase analysis results are abnormal, so that users can be informed.

7. The enterprise performance data control system according to claim 6, characterized in that, The result association and attribution analysis module (6) is also used to send linkage information to the process monitoring and attribution analysis module (7) when an abnormal assessment indicator is identified, so that the process monitoring and attribution analysis module (7) can confirm whether the process monitoring and attribution analysis module (7) has generated intervention reminder information corresponding to the abnormal assessment indicator within the assessment period corresponding to the abnormal assessment indicator. If so, the process monitoring and attribution analysis module (7) sends the phased analysis results of the process monitoring and attribution analysis module (7) when generating the corresponding intervention reminder information to the result association and attribution analysis module (6). The result association and attribution analysis module (6) is also used to, when receiving the interim analysis results, use the interim analysis results as an explanation of the abnormal assessment indicators and add them to the corresponding assessment results.

8. A method for controlling enterprise performance data, characterized in that, Raw performance data is obtained from multiple source systems within the enterprise, and the raw data is preprocessed. In response to user-defined requirements, the user-defined requirements are configured into a configuration file that can be parsed by a computer; wherein, the user-defined requirements include at least a user-defined assessment scheme; Obtain the configuration file and perform performance evaluation calculations according to the requirements of the configuration file to generate evaluation results; The assessment results are acquired and visualized so that users can be informed of them.

9. A device for controlling enterprise performance data, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described in claim 8.

10. A computer-readable storage medium, characterized in that, The computer program is stored that can be loaded by a processor and executed as described in claim 8.

Images