Enterprise assessment scheme generation method and device, electronic equipment and storage medium

Abstract

The invention discloses an enterprise assessment scheme generation method and device, electronic equipment and a storage medium, and relates to the technical field of artificial intelligence or other related fields, and the method comprises the steps: collecting the feature information of a target enterprise, and constructing a correlation model according to the feature information and a preset assessment index list, the association model is used for storing an association relationship between the feature item and the index item; according to the association model, matching and screening recommendation assessment indexes from a full-amount index library to form a recommendation index set; the recommendation index set is input into the target semantic reasoning model, an index weight distribution scheme is output, and each recommendation assessment index in the index weight distribution scheme corresponds to a recommendation weight value; and generating an enterprise assessment scheme corresponding to the target enterprise based on the recommended index set and the index weight distribution scheme. According to the invention, the technical problem of low scheme making efficiency caused by lack of unified intelligent tool support in the generation process of the enterprise assessment scheme in the prior art is solved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and more specifically, to a method and apparatus for generating enterprise performance evaluation schemes, electronic devices, and storage media. Background Technology

[0002] Traditional corporate performance evaluation schemes typically rely on manual experience and static templates, which have numerous limitations. In practice, strategists and senior managers need to manually identify corporate strategic goals and break them down into specific departmental and business line performance indicators. This process is time-consuming and highly subjective, easily influenced by human bias. For example, indicator selection may focus only on short-term financial goals while neglecting long-term market competitiveness or technological innovation capabilities; the allocation of weights may lack objective basis due to departmental competition. These problems result in a final performance evaluation scheme lacking scientific rigor and fairness, failing to effectively reflect the overall state of the company, and potentially even becoming disconnected from corporate strategy.

[0003] In existing technologies, some solutions employ expert-system-based performance evaluation tools, using pre-defined rule bases and logic trees to aid decision-making. However, these systems are typically closed and lack learning capabilities, making them ill-suited to rapidly changing market environments and corporate strategic adjustments. When significant business changes occur, the rule base requires extensive manual maintenance and updates, resulting in inefficiency. Other solutions focus on data-driven performance evaluation. However, these methods are often only applicable to quantitative indicators, optimizing through the analysis of historical data. They cannot effectively handle the selection and measurement of qualitative indicators (such as brand influence and customer satisfaction), nor can they fundamentally address the subjectivity issues in the decomposition of strategic goals and indicator selection.

[0004] In addition, there are some attempts to use knowledge graphs for enterprise information association and analysis. For example, some solutions use knowledge graphs to build internal enterprise knowledge bases for information retrieval and business association. However, these applications are mostly limited to the information query level and fail to delve into the generation and design of performance evaluation schemes, thus failing to address the issue of deep correlation between performance indicators and corporate strategic goals. Regarding weight design, there is a lack of ability to verify and optimize the overall logical rationality of the scheme, making it unable to handle complex business scenarios.

[0005] In summary, the relevant technologies suffer from a lack of unified intelligent tools to support the generation process of enterprise performance evaluation schemes, resulting in low efficiency in scheme development. Currently, no effective solution has been proposed to address these issues. Summary of the Invention

[0006] This invention provides a method, apparatus, electronic device, and storage medium for generating enterprise performance evaluation schemes, to at least solve the technical problem in the related art where the generation process of enterprise performance evaluation schemes lacks unified intelligent tool support, resulting in low efficiency in scheme formulation.

[0007] According to one aspect of the present invention, a method for generating an enterprise performance evaluation scheme is provided, comprising: collecting feature information of a target enterprise, and constructing an association model based on the feature information and a preset list of performance evaluation indicators, wherein the association model is used to store the association relationship between feature items and indicator items; matching and filtering N recommended performance evaluation indicators from a full indicator library based on the association model to form a recommended indicator set, wherein N is a positive integer; inputting the recommended indicator set into a target semantic reasoning model and outputting an indicator weight allocation scheme, wherein each recommended performance evaluation indicator in the indicator weight allocation scheme corresponds to a recommended weight value; and generating an enterprise performance evaluation scheme corresponding to the target enterprise based on the recommended indicator set and the indicator weight allocation scheme.

[0008] Further, the step of collecting characteristic information of the target enterprise and constructing an association model based on the characteristic information and a preset assessment indicator list includes: receiving enterprise identification information transmitted by the user through an interactive interface, wherein the enterprise identification information is used to identify the target enterprise; periodically calling a data acquisition engine to obtain data stream information, and extracting characteristic information of the target enterprise from the data stream information based on the enterprise identification information; constructing an enterprise profile of the target enterprise based on the characteristic information, wherein the enterprise profile is presented in a multimodal feature manner, the multimodality including: category attributes, numerical attributes, and text attributes; and constructing the association model corresponding to the target enterprise based on the preset assessment indicator list and the enterprise profile.

[0009] Further, the step of constructing the association model corresponding to the target enterprise based on the preset assessment indicator list and the enterprise profile includes: constructing enterprise feature nodes and assigning values ​​to the enterprise feature nodes based on the multimodal feature items in the enterprise profile to obtain first node information; constructing assessment indicator nodes and assigning values ​​to the assessment indicator nodes based on the preset assessment indicators in the preset assessment indicator list to obtain second node information; for each enterprise feature node and each assessment indicator node, using a similarity algorithm to calculate the similarity between the first node information and the second node information, and determining the similarity as an association weight value, wherein the magnitude of the association weight value is used to characterize the strength of the association between the feature item and the indicator item; establishing the association relationship between each enterprise feature node and each assessment indicator node based on the association weight value to obtain the association model corresponding to the target enterprise.

[0010] Furthermore, the full-scale indicator library is constructed through the following steps: extracting historical assessment schemes from the system database, wherein the historical assessment schemes cover multiple different enterprises, different development stages, and different business models; for each historical assessment scheme, analyzing the historical enterprise entities in the historical assessment scheme, the historical characteristic information possessed by the historical enterprise entities, the historical assessment indicators matching the historical characteristic information in the historical assessment scheme, and all the historical assessment indicators used in the historical assessment scheme; establishing a first mapping relationship between the historical enterprise entities and the historical characteristic information, establishing a second mapping relationship between the historical characteristic information and the historical assessment indicators, establishing a third mapping relationship between the historical assessment schemes and the historical assessment indicators, and establishing a fourth mapping relationship between the historical enterprise entities and the historical assessment schemes; and constructing the full-scale indicator library based on the first, second, third, and fourth mapping relationships.

[0011] Further, the step of matching and filtering N recommended assessment indicators from the full indicator library to form a recommended indicator set according to the association model includes: combining the association relationship indicated by the association model and the first mapping relationship in the full indicator library to determine historical feature information matching the target enterprise from the full indicator library; retrieving the enterprise profile of the target enterprise, filtering the historical feature information based on the multimodal features in the enterprise profile to obtain a first recommended feature set of the target enterprise; determining the historical assessment indicators corresponding to the recommended features from the full indicator library based on the recommended feature set and the second mapping relationship to obtain a first type of recommended indicators; if the full indicator library contains historical enterprise entities matching the target enterprise, determining the historical assessment scheme corresponding to the historical enterprise entity based on the fourth mapping relationship, and determining the historical assessment indicators involved in the historical assessment scheme based on the third mapping relationship to obtain a second type of recommended indicators; integrating the first type of recommended indicators and the second type of recommended indicators to obtain a recommended indicator set containing N recommended assessment indicators.

[0012] Further, the step of filtering the historical feature information based on the multimodal features in the enterprise profile to obtain the first recommended feature set of the target enterprise includes: extracting the multimodal features to obtain category attribute features, numerical attribute features, and text attribute features, and classifying the historical feature information into category attribute historical features, numerical attribute historical features, and text attribute historical features; using a cross-matching algorithm to calculate the cross-matching degree between the category attribute features of the target enterprise and each category attribute historical feature to obtain a first matching degree value; using a cosine similarity algorithm to calculate the cosine similarity between the numerical attribute features of the target enterprise and each numerical attribute historical feature to obtain a second matching degree value; using a vectorization algorithm to map the text attribute features of the target enterprise into text feature vectors, mapping the text attribute historical features into text historical feature vectors, and using the cosine similarity algorithm to calculate the cosine similarity between the text feature vector and each text historical feature vector to obtain a third matching degree value; and filtering the historical feature information corresponding to matching degree values ​​greater than or equal to a preset value to obtain the first recommended feature set of the target enterprise.

[0013] Further, the step of inputting the recommended indicator set into the target semantic reasoning model and outputting the indicator weight allocation scheme includes: using the target semantic reasoning model to call the auxiliary reasoning text of the target enterprise; performing semantic understanding on each recommended indicator in the recommended indicator set based on the auxiliary reasoning text to obtain semantic understanding information, wherein the auxiliary reasoning text is pre-stored in the database of the target semantic reasoning model, and the auxiliary reasoning text includes the target enterprise's corporate strategy information, industry characteristic information, and business model information; the semantic understanding information at least includes the impact of the recommended indicators on the corporate strategy, industry characteristics, and business model; using a weight allocation algorithm pre-installed in the target semantic reasoning model, combined with the semantic understanding information corresponding to all the recommended indicators, calculating the recommendation weight value corresponding to each recommended indicator to obtain the indicator weight allocation scheme, wherein the sum of the recommendation weight values ​​corresponding to all the recommended indicators is 1.

[0014] Furthermore, the step of generating the enterprise assessment scheme corresponding to the target enterprise based on the recommended indicator set and the indicator weight allocation scheme includes: calling a preset assessment process framework, constructing an initial assessment scheme corresponding to the target enterprise based on the preset assessment process framework; and integrating each recommended indicator in the recommended indicator set and its corresponding recommended weight value into the initial assessment scheme to obtain the enterprise assessment scheme corresponding to the target enterprise.

[0015] According to another aspect of the present invention, an apparatus for generating an enterprise performance evaluation scheme is also provided, comprising: a collection unit, configured to collect feature information of a target enterprise and construct an association model based on the feature information and a preset list of performance evaluation indicators, wherein the association model is used to store the association relationship between feature items and indicator items; a filtering unit, configured to match and filter N recommended performance evaluation indicators from a full indicator library according to the association model to form a recommended indicator set, wherein N is a positive integer; an inference unit, configured to input the recommended indicator set into a target semantic inference model and output an indicator weight allocation scheme, wherein each recommended performance evaluation indicator in the indicator weight allocation scheme corresponds to a recommended weight value; and a generation unit, configured to generate an enterprise performance evaluation scheme corresponding to the target enterprise based on the recommended indicator set and the indicator weight allocation scheme.

[0016] Further, the acquisition unit includes: a receiving module, used to receive enterprise identification information transmitted by the user through an interactive interface, wherein the enterprise identification information is used to identify the target enterprise; a first extraction module, used to periodically call the data acquisition engine to obtain data stream information, and extract feature information of the target enterprise from the data stream information based on the enterprise identification information; a first construction module, used to construct an enterprise profile of the target enterprise based on the feature information, wherein the enterprise profile is presented in the form of multimodal features, the multimodality including: category attributes, numerical attributes and text attributes; and a second construction module, used to construct the association model corresponding to the target enterprise based on the preset assessment indicator list and the enterprise profile.

[0017] Further, the second construction module includes: a first construction submodule, used to construct enterprise feature nodes and assign values ​​to the enterprise feature nodes based on the multimodal feature items in the enterprise profile to obtain first node information; a second construction submodule, used to construct assessment indicator nodes and assign values ​​to the assessment indicator nodes based on the preset assessment indicators in the preset assessment indicator list to obtain second node information; a first calculation submodule, used to calculate the similarity between the first node information and the second node information for each enterprise feature node and each assessment indicator node using a similarity algorithm, and determine the similarity as an association weight value, wherein the magnitude of the association weight value is used to characterize the strength of the association relationship between the feature item and the indicator item; and an establishment submodule, used to establish the association relationship between each enterprise feature node and each assessment indicator node based on the association weight value to obtain the association model corresponding to the target enterprise.

[0018] Furthermore, the enterprise performance evaluation scheme generation device further includes: a construction unit for constructing the full-scale indicator library, the construction unit comprising: a second extraction module for extracting historical performance evaluation schemes from the system database, wherein the historical performance evaluation schemes cover multiple different enterprises, different development stages, and different business models; an analysis module for analyzing, for each historical performance evaluation scheme, the historical enterprise entities in the historical performance evaluation scheme, the historical characteristic information possessed by the historical enterprise entities, the historical performance evaluation indicators matching the historical characteristic information in the historical performance evaluation scheme, and all the historical performance evaluation indicators used in the historical performance evaluation scheme; an establishment module for establishing a first mapping relationship between the historical enterprise entities and the historical characteristic information, a second mapping relationship between the historical characteristic information and the historical performance evaluation indicators, a third mapping relationship between the historical performance evaluation schemes and the historical performance evaluation indicators, and a fourth mapping relationship between the historical enterprise entities and the historical performance evaluation schemes; and a third construction module for constructing the full-scale indicator library based on the first mapping relationship, the second mapping relationship, the third mapping relationship, and the fourth mapping relationship.

[0019] Further, the filtering unit includes: a first determining module, used to combine the association relationship indicated by the association model and the first mapping relationship in the full indicator library to determine historical feature information matching the target enterprise from the full indicator library; a filtering module, used to retrieve the enterprise profile of the target enterprise, and filter the historical feature information based on the multimodal features in the enterprise profile to obtain a first recommended feature set of the target enterprise; a second determining module, used to determine the historical assessment indicators corresponding to the recommended features from the full indicator library based on the recommended feature set and the second mapping relationship to obtain a first type of recommended indicators; a third determining module, used to determine the historical assessment scheme corresponding to the historical enterprise entity based on the fourth mapping relationship when the full indicator library contains the historical enterprise entity matching the target enterprise, and to determine the historical assessment indicators involved in the historical assessment scheme based on the third mapping relationship to obtain a second type of recommended indicators; and a first integrating module, used to integrate the first type of recommended indicators and the second type of recommended indicators to obtain a recommended indicator set containing N recommended assessment indicators.

[0020] Further, the filtering module includes: an extraction submodule, used to extract the multimodal features to obtain category attribute features, numerical attribute features, and text attribute features, and classify the historical feature information into category attribute historical features, numerical attribute historical features, and text attribute historical features; a second calculation submodule, used to calculate the cross-matching degree between the category attribute features of the target enterprise and each category attribute historical feature using a cross-matching algorithm to obtain a first matching degree value; a third calculation submodule, used to calculate the cosine similarity between the numerical attribute features of the target enterprise and each numerical attribute historical feature using a cosine similarity algorithm to obtain a second matching degree value; a fourth calculation submodule, used to map the text attribute features of the target enterprise into text feature vectors and the text attribute historical features into text historical feature vectors using a vectorization algorithm, and calculate the cosine similarity between the text feature vector and each text historical feature vector using the cosine similarity algorithm to obtain a third matching degree value; and a filtering submodule, used to filter the historical feature information corresponding to matching degree values ​​greater than or equal to a preset value to obtain a first recommended feature set of the target enterprise.

[0021] Further, the reasoning unit includes: a semantic understanding module, used to call the auxiliary reasoning text of the target enterprise using the target semantic reasoning model, and perform semantic understanding on each recommendation indicator in the recommendation indicator set based on the auxiliary reasoning text to obtain semantic understanding information, wherein the auxiliary reasoning text is pre-stored in the database of the target semantic reasoning model, and the auxiliary reasoning text includes the target enterprise's corporate strategy information, industry characteristic information, and business model information, and the semantic understanding information at least includes the impact of the recommendation indicators on the corporate strategy, industry characteristics, and business model; and a calculation module, used to use a weight allocation algorithm pre-installed in the target semantic reasoning model, combined with the semantic understanding information corresponding to all the recommendation indicators, to calculate the recommendation weight value corresponding to each recommendation indicator, and obtain the indicator weight allocation scheme, wherein the sum of the recommendation weight values ​​corresponding to all the recommendation indicators is 1.

[0022] Furthermore, the generation unit includes: a fourth construction module, used to call a preset assessment process framework and construct an initial assessment scheme corresponding to the target enterprise based on the preset assessment process framework; and a second integration module, used to integrate each of the recommended indicators in the recommended indicator set and each of the corresponding recommended weight values ​​into the initial assessment scheme to obtain the enterprise assessment scheme corresponding to the target enterprise.

[0023] According to another aspect of the present invention, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute any of the above-described methods for generating enterprise assessment schemes.

[0024] According to another aspect of the present invention, an electronic device is also provided, including one or more processors and a memory, the memory being used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement any of the above-described methods for generating enterprise assessment schemes.

[0025] This invention proposes a method for generating enterprise performance evaluation schemes. First, characteristic information of the target enterprise is collected. Then, based on the characteristic information and a pre-set list of performance indicators, an association model is constructed. This association model stores the relationships between characteristic items and indicator items. Next, based on the association model, N recommended performance indicators are matched and selected from the full indicator library to form a recommended indicator set, where N is a positive integer. The recommended indicator set is then input into the target semantic reasoning model, which outputs an indicator weight allocation scheme. In this scheme, each recommended performance indicator corresponds to a recommended weight value. Finally, based on the recommended indicator set and the indicator weight allocation scheme, the enterprise performance evaluation scheme corresponding to the target enterprise is generated.

[0026] This invention employs a combination of deep learning and graph analysis. By constructing precise corporate profiles and intelligently selecting assessment indicators, it achieves the goal of automatically generating scientific and personalized corporate assessment schemes. This improves the efficiency and accuracy of scheme formulation and solves the technical problem in related technologies where the corporate assessment scheme generation process lacks unified intelligent tools, resulting in low scheme formulation efficiency, reliance on subjective experience, and difficulty in coping with complex and ever-changing corporate environments and strategic needs.

[0027] Specifically, this invention first utilizes data acquisition and processing technologies to extract key feature information from multi-source data of the target enterprise, constructing a multimodal enterprise profile that comprehensively reflects the enterprise's current status and strategic orientation. Then, based on a highly structured association model, combined with the enterprise profile and a pre-set list of performance indicators, it quickly locates and filters a set of recommended performance indicators that highly match the enterprise's characteristics from the full indicator library. This process not only considers the direct correlation between indicators and enterprise characteristics but also utilizes historical data and the experience of similar enterprises to ensure the scientific and rational selection of indicators. Next, the selected set of recommended indicators is input into the target semantic reasoning model. Through deep semantic understanding and logical reasoning, an indicator weight allocation scheme is output. This scheme not only reflects the relative importance of indicators in enterprise performance evaluation but also considers the complex influence of enterprise strategy, industry characteristics, and business models, thereby avoiding subjectivity and arbitrariness in weight allocation and ensuring the objectivity and fairness of the evaluation scheme. Finally, based on the recommended indicator set and the indicator weight allocation scheme, a complete and accurate enterprise performance evaluation scheme is generated. This scheme not only accurately reflects the enterprise's strategic goals but also adapts to the dynamic changes in enterprise development, providing strong technical support for enterprise performance management and strategic execution. The above steps effectively improve the efficiency and quality of enterprise performance evaluation scheme development, realize the intelligentization of the entire process from scheme development to implementation, solve the problems of low efficiency, strong subjectivity and lack of scientific basis in the traditional scheme development process, and further solve the technical problem of low efficiency in scheme development caused by the lack of unified intelligent tools to support the enterprise performance evaluation scheme generation process. Attached Figure Description

[0028] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0029] Figure 1 This is a flowchart of an optional enterprise assessment scheme intelligent recommendation process according to an embodiment of the present invention;

[0030] Figure 2 This is a flowchart of an optional enterprise performance evaluation scheme generation method according to an embodiment of the present invention;

[0031] Figure 3 This is a module diagram of an optional intelligent recommendation system for enterprise performance evaluation schemes according to an embodiment of the present invention;

[0032] Figure 4 This is a schematic diagram of an optional enterprise assessment scheme generation device according to an embodiment of the present invention;

[0033] Figure 5 This is a hardware structure block diagram of an electronic device (or mobile device) that performs a method for generating an enterprise assessment scheme according to an embodiment of the present invention. Detailed Implementation

[0034] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0035] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0036] To facilitate understanding of the present invention by those skilled in the art, some terms or nouns involved in the various embodiments of the present invention are explained below:

[0037] The Full Metric Library is a pre-built, structured, and scalable set of performance indicators by corporate strategy experts, based on different industries, company types, business models, and other dimensions. This library serves as the candidate source for the intelligent selection of performance indicators in this invention; it is a conceptual, structured knowledge system, rather than a simple list of indicators.

[0038] A knowledge graph is a structured method for representing knowledge, used to describe entities, concepts, and their relationships in the objective world. In this invention, it specifically refers to a knowledge graph used to construct enterprise profiles, where nodes are enterprise entities (such as enterprises, industries, and products), profile features, indicators, and relationships (such as "the enterprise has a certain feature" or "the indicator is applicable to a certain feature").

[0039] Enterprise profiling, based on knowledge graph technology, is a multi-dimensional digital description formed by deeply linking and modeling various aspects of information within an enterprise.

[0040] LLM, or Large Language Model, refers to a deep learning model trained with a large number of parameters and massive amounts of text data, possessing powerful capabilities in natural language understanding, logical reasoning, text generation, and knowledge integration. In this invention, LLM acts as an intelligent agent, responsible for metric optimization and weight verification.

[0041] AHP, or Analytic Hierarchy Process, is a systematic and hierarchical decision-making method that combines qualitative and quantitative analysis to solve weight allocation problems in complex decision-making.

[0042] Back Propagation (BP) is a multi-layer feedforward neural network whose core feature is the use of the back propagation algorithm for supervised learning.

[0043] The following embodiments of the present invention can be applied to various systems / applications / devices that require the construction and weight allocation of enterprise performance appraisal indicator systems, enabling intelligent appraisal scheme generation based on knowledge graphs and large-scale language models. Technically, the present invention divides the enterprise appraisal scheme generation process into several key modules, including data collection and enterprise profile construction, intelligent indicator matching, scientific weight calculation and intelligent verification. An efficient information processing channel is pre-established between the enterprise profile and the large-scale language model using a deep learning architecture, facilitating intelligent correlation analysis between the enterprise profile and the full indicator database, and accurate recommendation of appraisal indicators.

[0044] The present invention will now be described in detail with reference to various embodiments.

[0045] Example 1

[0046] According to an embodiment of the present invention, a method for generating an enterprise assessment scheme is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0047] Examples of embodiments of the present invention Figure 2The method for generating enterprise performance appraisal schemes, as shown, is implemented by an intelligent enterprise performance appraisal recommendation system. This system combines knowledge graph and large-scale language model technologies to automatically generate enterprise performance appraisal schemes. Specifically, it addresses the problems of low efficiency, strong subjectivity, and lack of scientific basis in traditional scheme development. This is achieved through constructing a refined enterprise profile, intelligently selecting performance indicators, and scientifically calculating indicator weights. The process involves data collection and profile construction, indicator selection and calculation, weight allocation, and scheme generation steps to generate scientific and personalized enterprise performance appraisal schemes. This series of innovative technological processes not only improves the efficiency and quality of performance appraisal scheme development but also ensures that the schemes can comprehensively and accurately evaluate the overall performance of the enterprise, promoting the modernization and intelligent development of the performance management system.

[0048] Figure 1 This is a flowchart of an optional intelligent recommendation process for enterprise performance evaluation according to an embodiment of the present invention, such as... Figure 1 As shown, this process combines expert knowledge with artificial intelligence technology to achieve intelligent optimization of assessment indicators and scientific verification of weight design through a systematic process, thereby generating a scientific, efficient assessment plan that aligns with corporate strategy. The overall process begins with the user selecting a target company in the browser. Then, the system enters the intelligent indicator selection layer. Utilizing the multi-dimensional relationships and graph query mechanism of a knowledge graph (KG), the system performs initial screening and pruning based on the key assessment elements of the company profile, quickly obtaining a set of candidate indicators highly relevant to the company's strategy. Next, this module combines the powerful semantic reasoning capabilities of a general large model (LLM) to conduct in-depth analysis and evaluation of the initial screening results, intelligently optimizing the indicator set. This set is then manually confirmed and adjusted to become the final indicator set. The data preparation layer relies on Apache Flink (streaming data) to integrate and process multi-source data from external systems and internal documents, generating a full indicator library. Structured indicators are stored in a database (e.g., MongoDB), and the company profile and its relationships are persisted to a graph database (e.g., Neo4j). Finally, in the indicator weight calculation layer, the weights of each indicator are calculated based on the AHP algorithm, and then optimized and calibrated by a general large model to output a scientific and reasonable enterprise assessment recommendation scheme. After manual confirmation and adjustment, it becomes the final assessment scheme.

[0049] Figure 2 This is a flowchart of an optional enterprise performance evaluation scheme generation method according to an embodiment of the present invention, such as... Figure 2 As shown, the method includes the following steps:

[0050] Step S201: Collect the characteristic information of the target enterprise, and construct an association model based on the characteristic information and the preset assessment indicator list. The association model is used to store the relationship between characteristic items and indicator items.

[0051] Specifically, the target enterprise refers to a specific enterprise that receives the data generated by the intelligent assessment scheme. In this embodiment of the invention, the intelligent assessment scheme utilizes not only a static dataset, but also dynamically captures multi-dimensional information such as the enterprise's development stage, industry characteristics, and business pain points through the target enterprise's corporate profile, thereby providing rich background data for the generation of the assessment scheme. The target enterprise's characteristic information covers discrete attributes (such as industry and development stage), continuous attributes (such as scale characteristics), and textual attributes (such as strategic focus), originating not only from the enterprise's internal database but also from publicly available external data or information integrated from third-party systems, ensuring the comprehensiveness of the corporate profile.

[0052] It's important to note that the preset performance indicator list is a pre-constructed set of structured indicators, encompassing various dimensions of performance indicators that different companies may have, such as strategy, finance, marketing, and human resources. It serves as the initial knowledge base and candidate pool for intelligent indicator selection. The association model, built using knowledge graph technology, stores the relationships between feature items and indicator items. Through the connection of enterprise feature nodes and performance indicator nodes, and the assignment of weights, it forms a graph structure that enables efficient querying and analysis of indicator applicability. For example, if the target company is in a transformation period and its strategic focus is on technological innovation, the association model will prioritize recommending performance indicators related to "transformation period" and "technological innovation."

[0053] The process of storing the relationships between features and indicators involves leveraging the characteristics of graph databases to structurally store the logical connections between enterprise features and performance indicators in the form of nodes and edges. This approach not only visually displays the relationships between indicators and enterprise features but also utilizes the query capabilities of graph databases and the reasoning capabilities of large-scale language models to quickly locate the most suitable performance indicators, thereby enabling intelligent generation of performance evaluation schemes.

[0054] Further, the steps of collecting characteristic information of the target enterprise and constructing an association model based on the characteristic information and a pre-set list of assessment indicators include: receiving enterprise identification information transmitted by the user through an interactive interface, wherein the enterprise identification information is used to identify the target enterprise; periodically calling the data acquisition engine to obtain data stream information, and extracting the characteristic information of the target enterprise from the data stream information based on the enterprise identification information; constructing an enterprise profile of the target enterprise based on the characteristic information, wherein the enterprise profile is presented in the form of multimodal features, including: category attributes, numerical attributes, and text attributes; and constructing an association model corresponding to the target enterprise based on the pre-set list of assessment indicators and the enterprise profile.

[0055] It should be noted that enterprise identification information refers to the unique code or identifier used internally by the system to identify and locate target enterprises. It is the starting point for data collection and enterprise profile construction. Multimodal features refer to the various types of features included in the enterprise profile, including categorical attributes (discrete labels), numerical attributes (quantitative indicators), and textual attributes (unstructured text descriptions), used to comprehensively depict the current status of the enterprise. The association model refers to a model built based on knowledge graph technology to represent the relationship between enterprise characteristics and performance indicators. It is stored in a graph structure for efficient querying and analysis.

[0056] In this embodiment of the invention, collecting characteristic information of the target enterprise and constructing an association model based on this information and a preset list of assessment indicators can be broken down into the following specific implementation steps: The user inputs a specific enterprise identifier (e.g., enterprise ID, unified social credit code, etc.) through the interactive interface provided by the system (e.g., web form or mobile application interface). This identifier is used by the system to locate and extract relevant information of the enterprise; the system periodically (e.g., daily, weekly) calls a data acquisition engine (e.g., Apache Flink) to obtain real-time data streams from multi-source data (including but not limited to internal enterprise databases, external public data, industry reports, etc.). These data streams contain information on the enterprise, finance, human resources, and other aspects, and are important data sources for constructing enterprise profiles; based on the characteristic information extracted from the data streams, the system constructs an enterprise profile of the target enterprise, including category attributes (e.g., industry, development stage), numerical attributes (e.g., size, profitability), and textual attributes (e.g., strategic planning, market feedback). The construction of enterprise profiles is based on knowledge graph technology. Multi-dimensional data is linked and structured through graph databases, providing a solid foundation for subsequent intelligent analysis. Combining the pre-set list of assessment indicators and the enterprise profiles constructed above, a corresponding association model for the target enterprise is built. This model can be stored in the form of a graph, where nodes represent enterprise characteristics and assessment indicators, and edges represent the relationship between the two, including but not limited to "applicability" and "importance". The graph query algorithm can quickly locate assessment indicators that match the enterprise characteristics.

[0057] Furthermore, the steps for constructing the association model corresponding to the target enterprise based on the preset list of assessment indicators and the enterprise profile include: constructing enterprise feature nodes and assigning values ​​to the enterprise feature nodes based on the multimodal feature items in the enterprise profile to obtain the first node information; constructing assessment indicator nodes and assigning values ​​to the assessment indicator nodes based on the preset assessment indicators in the preset list of assessment indicators to obtain the second node information; for each enterprise feature node and each assessment indicator node, using a similarity algorithm, calculating the similarity between the first node information and the second node information, and determining the similarity as the association weight value, wherein the magnitude of the association weight value is used to characterize the strength of the association between the feature item and the indicator item; establishing the association relationship between each enterprise feature node and each assessment indicator node based on the association weight value to obtain the association model corresponding to the target enterprise.

[0058] Specifically, the first node information refers to the enterprise feature node information, which consists of specific feature values ​​from the enterprise profile and is a key component of the graph database. The second node information refers to the performance indicator node information, which is based on specific indicators from a pre-set list of performance indicators and is used for matching analysis with the enterprise feature nodes. Similarity algorithms are used to quantify the degree of similarity between two nodes or information sets; for example, the similarity coefficient is used to compare discrete features, and cosine similarity is used to compare normalized continuous and textual attributes. The association weight value represents the quantified value of the association strength between the enterprise feature node and the performance indicator node; the magnitude of the weight value directly determines the priority of the indicator in the intelligent screening process.

[0059] It's important to note that constructing enterprise feature nodes refers to converting a company's discrete attributes (such as industry and development stage), continuous attributes (such as size characteristics), and textual attributes (such as strategic focus) into node information in a graph database. Each attribute becomes a node in the graph, called an enterprise feature node. Based on enterprise profile data, branch nodes can be created with the enterprise ID as the root node, and each branch node represents an enterprise feature. Examples include "industry feature" nodes and "size feature" nodes.

[0060] Constructing performance indicator nodes refers to creating corresponding nodes based on the various indicator information covered in the pre-set performance indicator list. Each node records detailed indicator attributes, such as indicator ID, name, target range, and calculation logic. Each indicator in the indicator library can be entered into a graph database as a node, with each node accompanied by corresponding metadata to ensure matching and retrieval with enterprise characteristic nodes.

[0061] Calculating association weight values ​​using similarity algorithms refers to determining the strength of the association between enterprise feature nodes and performance indicator nodes, which guides subsequent indicator selection and weight allocation. Algorithms such as similarity coefficient (for discrete attributes) and cosine similarity (for normalized continuous and textual attributes) can be used to quantify the degree of matching between enterprise features and indicators. The calculation results are used as association weight values ​​to reflect the tightness of the association between the two.

[0062] Establishing relationships refers to constructing a network of relationships between enterprise feature nodes and performance indicator nodes in a graph database, forming a relationship model. Based on the calculated relationship weights, the attribute relationships of the graph database can be used to connect enterprise feature nodes and performance indicator nodes. By setting the relationship type (such as APPLICABLE_TO or HAS_FEATURE) and weight attributes, it can be ensured that the model can intuitively reflect the applicability and importance of the indicators and enterprise features.

[0063] The above steps aim to achieve intelligent generation and personalized weight allocation of corporate performance indicators. The technical goal is to automatically generate performance evaluation schemes that align with corporate strategy and are highly operable by deeply understanding and matching corporate characteristics with performance indicators, thereby improving the efficiency and scientific nature of corporate management decisions.

[0064] Step S202: Based on the association model, N recommended assessment indicators are matched and selected from the full indicator library to form a recommended indicator set, where N is a positive integer.

[0065] It should be noted that the pre-built full-scale indicator library contains a set of all possible performance indicators, covering various dimensions such as finance, marketing, and human resources. Each indicator is accompanied by detailed definitions, calculation logic, data sources, and applicable scenarios. The full-scale indicator library is the initial knowledge base for intelligent indicator selection and weight calculation in this invention, and indicators can be classified according to dimensions such as strategy, finance, customers, and innovation using a hierarchical structure.

[0066] The recommended performance indicators are those that best match the characteristics of the enterprise, selected from the full indicator library based on enterprise profiles and correlation models. These indicators are intelligently optimized through multimodal similarity calculation, comparison with high-frequency indicators in similar historical enterprise solutions, and deep semantic understanding using large-scale language models. They not only align with the characteristics of the enterprise at the quantitative level, but also highly align with the enterprise's strategic goals and pain point analysis at the semantic level, making them the core elements for forming the final performance evaluation scheme.

[0067] The above steps, through intelligent means, efficiently and accurately select the most suitable performance indicators for the target company from the full indicator database, forming a personalized recommended indicator set. This addresses the problems of low efficiency, strong subjectivity, and lack of scientific basis in the traditional performance indicator selection process. This process improves the automation level of performance indicator selection while ensuring a close match between the indicators and the company's characteristics and strategic goals.

[0068] Furthermore, the full-scale indicator library is constructed through the following steps: historical assessment schemes are extracted from the system database, covering multiple different enterprises, different development stages, and different business models; for each historical assessment scheme, the historical enterprise entities, their historical characteristic information, historical assessment indicators matching the historical characteristic information, and all historical assessment indicators used in the historical assessment scheme are analyzed; a first mapping relationship is established between historical enterprise entities and historical characteristic information, a second mapping relationship is established between historical characteristic information and historical assessment indicators, a third mapping relationship is established between historical assessment schemes and historical assessment indicators, and a fourth mapping relationship is established between historical enterprise entities and historical assessment schemes; based on the first, second, third, and fourth mapping relationships, the full-scale indicator library is constructed.

[0069] It should be noted that various previously implemented performance evaluation schemes, i.e., historical performance evaluation schemes, are extracted from the system database. These schemes cover different enterprises, different stages of development, and different business models, providing a rich and diverse data foundation for building a comprehensive indicator library. For each historical performance evaluation scheme, the relevant enterprise entities (historical enterprises) are analyzed in depth, and their characteristic information (such as industry, development stage, pain points, etc.) and the performance indicators that match these characteristics in the historical performance evaluation schemes are extracted, thus establishing a link between enterprise characteristics and performance indicators.

[0070] The first mapping relationship establishes the connection between historical enterprise entities and their respective characteristic information, ensuring that the characteristic information of each enterprise entity can be accurately tracked and referenced. The second mapping relationship is constructed by mapping historical characteristic information to historical performance indicators, clarifying which indicators are typically associated with specific enterprise characteristics, providing a basis for further indicator selection and recommendation. The third mapping relationship is established by mapping historical performance evaluation schemes to historical performance indicators, recording all specific indicators involved in the schemes, ensuring the reusability and consistency of indicators across different schemes. The fourth mapping relationship establishes the connection between historical enterprise entities and historical performance evaluation schemes, reflecting how enterprises designed and implemented performance evaluation schemes at different development stages and under different business models, providing current enterprises with referable cases. Finally, based on the above mapping relationships, a comprehensive indicator library is constructed through aggregation, organization, and optimization. This library not only includes all historically used performance indicators but also adds information on the correlation between indicators and enterprise characteristics and performance evaluation schemes, providing a comprehensive and structured knowledge base for intelligent performance indicator selection and weight allocation.

[0071] The goal of the above steps is to achieve intelligent generation and dynamic optimization of the enterprise performance indicator system. By deeply understanding and matching enterprise characteristics with performance indicators, and combining historical data and industry knowledge, the system automatically generates performance plans that are both in line with the enterprise strategy and highly operable. This improves the efficiency and scientific nature of enterprise management decisions and ensures the consistency and effectiveness of performance management with the enterprise's strategic goals.

[0072] Further, based on the association model, the step of matching and selecting N recommended assessment indicators from the full indicator library to form a recommended indicator set includes: combining the association relationship indicated by the association model and the first mapping relationship in the full indicator library to determine the historical feature information matching the target enterprise from the full indicator library; retrieving the enterprise profile of the target enterprise, filtering the historical feature information based on the multimodal features in the enterprise profile to obtain the first recommended feature set of the target enterprise; based on the recommended feature set and the second mapping relationship, determining the historical assessment indicators corresponding to the recommended features from the full indicator library to obtain the first type of recommended indicators; if the full indicator library contains historical enterprise entities matching the target enterprise, determining the historical assessment scheme corresponding to the historical enterprise entity based on the fourth mapping relationship, and determining the historical assessment indicators involved in the historical assessment scheme based on the third mapping relationship to obtain the second type of recommended indicators; integrating the first type of recommended indicators and the second type of recommended indicators to obtain a recommended indicator set containing N recommended assessment indicators.

[0073] It should be noted that, combined with the guidance of the association model, and based on the first mapping relationship in the full indicator library, locating historical feature information that matches the characteristics of the target enterprise can be achieved through graph query algorithms in graph databases, ensuring that the most relevant historical data to the current enterprise can be captured. For example, by querying the knowledge graph, historical feature information that matches the target enterprise's industry, development stage, or specific business model can be found, providing guidance for subsequent indicator selection. Based on the target enterprise's profile (including multimodal features), further filtering is performed from the historical feature information to form a first recommended feature set that is more closely matched to the current enterprise. This can be achieved by using the category attributes, numerical attributes, and text attributes in the enterprise profile to perform multi-dimensional comparisons with historical feature information, selecting the feature set most similar to the target enterprise. Based on the first recommended feature set and the second mapping relationship in the full indicator library, the first type of recommended indicators that are highly relevant to the recommended feature set are selected from the historical assessment indicators. This can be achieved through graph database queries and matching, combined with the applicability tags between indicators and features (such as APPLICABLE_TO), to select the set of indicators with the highest matching degree to the first recommended feature set from the full indicator library.

[0074] In addition, historical enterprise entities with high similarity to the target enterprise are found in the full indicator database. Their historical assessment schemes are located based on the fourth mapping relationship. Then, the historical assessment indicators used in these schemes are determined through the third mapping relationship. As the second type of recommended indicators, the historical enterprise ID with the highest feature similarity to the target enterprise can be used. The historical assessment schemes corresponding to the historical enterprise entities are queried through the graph database. The specific indicators used in the schemes are further extracted and recommended to the current enterprise as practical cases.

[0075] The first type of recommendation indicators, selected based on historical feature information, and the second type of recommendation indicators, determined based on historical enterprise entity plans, are integrated to eliminate duplication and form a set of recommendation indicators that includes the recommendation assessment indicators. Specifically, this means performing a union operation on the two types of recommendation indicators, removing duplicate indicators, and finally selecting the most relevant indicators to form a personalized set of recommendation indicators.

[0076] The above steps enable the intelligent generation and personalized matching of corporate performance indicator systems. By constructing an efficient correlation model between target companies and performance indicators, and combining historical data with an expert-defined indicator library, the system intelligently selects the combination of performance indicators that best matches the company's current situation and strategic goals. This provides a precise basis for subsequent weight calculations and the generation of performance evaluation schemes, thereby improving the scientific nature and efficiency of corporate performance management.

[0077] Further, the step of filtering historical feature information based on multimodal features in the enterprise profile to obtain the first recommended feature set of the target enterprise includes: extracting multimodal features to obtain categorical attribute features, numerical attribute features, and textual attribute features, and classifying historical feature information into categorical attribute historical features, numerical attribute historical features, and textual attribute historical features; using a cross-matching algorithm to calculate the cross-matching degree between the categorical attribute features of the target enterprise and each categorical attribute historical feature to obtain a first matching degree value; using a cosine similarity algorithm to calculate the cosine similarity between the numerical attribute features of the target enterprise and each numerical attribute historical feature to obtain a second matching degree value; using a vectorization algorithm to map the textual attribute features of the target enterprise into textual feature vectors, and mapping the textual attribute historical features into textual historical feature vectors, and using a cosine similarity algorithm to calculate the cosine similarity between the textual feature vector and each textual historical feature vector to obtain a third matching degree value; and filtering historical feature information corresponding to matching degree values ​​greater than or equal to a preset value to obtain the first recommended feature set of the target enterprise.

[0078] It's important to note that extracting categorical attributes (such as industry and development stage), numerical attributes (such as company size and profitability), and textual attributes (such as strategic planning and market feedback) from the target company profile allows for the use of NLP (Natural Language Processing) techniques to process and extract features from unstructured text, converting it into quantifiable vector representations. Simultaneously, structured data, such as industry type and financial ratios, can be obtained from databases as categorical and numerical attribute features. For example, the system might identify the target company's industry as "software services," its development stage as "growth stage," and its main operational pain point as "high customer churn rate"—these are all categorical attribute features. Company size (e.g., 500 employees) and revenue ratio are numerical attribute features. Furthermore, analyzing strategic text can extract textual attribute features such as "improving customer satisfaction."

[0079] The goal of calculating the first matching score (category attribute) is to quantify the degree of matching between the category attribute features of the target company and the category attributes in historical feature information. Algorithms such as Jaccard similarity and Euclidean distance can be used to compare the discrete attributes in the target company and historical companies and calculate the similarity. For example, the Jaccard similarity of the "industry type" and "development stage" features between the target company and historical companies can be calculated, and the first matching score can be output.

[0080] The goal of calculating the second matching degree (numerical attribute) is to quantify the similarity between the numerical attribute features of the target company and the numerical attributes in historical feature information. This can be achieved by normalizing the numerical features and then using a cosine similarity algorithm to calculate the matching degree. For example, the "revenue growth rate" of the target company can be normalized and its corresponding historical indicators can be normalized, and then the cosine similarity can be calculated as the second matching degree.

[0081] The goal of calculating the third matching score (text attributes) is to assess the semantic similarity between the text attribute features of the target company and the features of historical texts. A pre-trained text embedding model (such as BERT) can be used to convert the text attributes into vectors, and then the cosine similarity can be calculated. For example, the strategic planning text of the target company can be converted into a vector representation, and the cosine similarity between this vector and the historical strategic texts of other companies can be calculated to obtain the third matching score.

[0082] By combining the first, second, and third matching scores, the historical feature information most similar to the target company can be filtered out to form the first recommended feature set. A preset matching score threshold, such as 80%, can be set. From all the calculated matching score values, feature information greater than or equal to this threshold is selected to form the recommendation set. For example, the system ultimately retains those historical feature information with a high matching score (e.g., greater than 80%) with the target company. This information constitutes the first recommended feature set and is used as the basis for subsequent indicator selection.

[0083] The above steps enable intelligent and personalized recommendation and construction of enterprise performance indicators. Through in-depth mining and matching of multimodal features, combined with historical data and industry knowledge, the system automatically generates performance indicator recommendations that best match the enterprise's current situation and strategic goals, thereby improving the efficiency and scientific nature of enterprise performance management.

[0084] Step S203: Input the set of recommended indicators into the target semantic reasoning model and output the indicator weight allocation scheme, wherein each recommended assessment indicator in the indicator weight allocation scheme corresponds to a recommended weight value.

[0085] The target semantic reasoning model in this invention refers to a large-scale language model (LLM). Based on deep understanding and reasoning capabilities, it comprehensively evaluates the input set of recommendation indicators and the company's strategic text, business situation, etc., and outputs a weight allocation scheme for the indicators. This model is trained on a massive corpus and can understand the meaning in complex contexts, process unstructured information, and make reasonable logical judgments. In this embodiment, LLM can not only allocate indicator weights according to the company's characteristics, but also perform deep semantic matching between the indicators and the company's strategic goals, ensuring that the allocated weights are both scientific and consistent with the company's actual situation.

[0086] It's important to note that the indicator weighting scheme is a set of weight values ​​output by the target semantic reasoning model, assigned to each recommended performance indicator. This reflects the relative importance of each indicator within the overall performance evaluation scheme. A reasonable allocation of weight values ​​helps companies more accurately measure performance and identify key performance indicators. Based on the characteristics and strategic needs of the target company, the weighting scheme uses intelligent analysis through LLM to assign weights to each recommended performance indicator, guiding the company on how to allocate resources and attention among different performance indicators to achieve more effective performance evaluation. The recommended weight value is the specific numerical value corresponding to each recommended performance indicator in the indicator weighting scheme, representing the importance of that indicator in the overall performance evaluation.

[0087] The steps described in this invention intelligently allocate indicator weights by introducing a target semantic reasoning model (such as an LLM based on the Transformer architecture), thus solving the problems of subjectivity, lack of scientific basis, and mismatch with corporate strategy in the traditional assessment scheme formulation process. By combining corporate profiles with historical data, LLM can provide a weight allocation scheme that reflects both corporate strategy and business realities.

[0088] Further, the steps of inputting the recommended indicator set into the target semantic reasoning model and outputting the indicator weight allocation scheme include: using the target semantic reasoning model to call the auxiliary reasoning text of the target enterprise, performing semantic understanding on each recommended indicator in the recommended indicator set based on the auxiliary reasoning text to obtain semantic understanding information. The auxiliary reasoning text is pre-stored in the database of the target semantic reasoning model and includes the target enterprise's corporate strategy information, industry characteristic information, and business model information. The semantic understanding information at least includes the impact of the recommended indicators on the corporate strategy, industry characteristics, and business model. Using a weight allocation algorithm pre-built in the target semantic reasoning model, combined with the semantic understanding information corresponding to all recommended indicators, the recommended weight value corresponding to each recommended indicator is calculated to obtain the indicator weight allocation scheme. The sum of the recommended weight values ​​corresponding to all recommended indicators is 1.

[0089] It should be noted that the auxiliary reasoning text is a collection of texts pre-stored in the target semantic reasoning model database. It contains key information such as the target company's strategic information, industry characteristics, and business model. It is used to help the model understand the company's background and generate more personalized and targeted recommendations.

[0090] This invention retrieves auxiliary reasoning text for the target company from a database associated with the target semantic reasoning model. This text includes corporate strategy information, industry characteristic information, and business model information. The text in the database has undergone preprocessing, such as text cleaning, word segmentation, and part-of-speech tagging, and is stored in a structured or vectorized form to facilitate model understanding and analysis.

[0091] In one alternative embodiment, if the target enterprise is a software service provider, its auxiliary reasoning text includes content such as "digital transformation strategy," "Software as a Service (SaaS) business model," and "cloud computing industry trends." This text will serve as key input for the model to understand the enterprise's background and strategic needs.

[0092] Furthermore, the target semantic reasoning model performs deep semantic understanding on each indicator in the recommended indicator set, outputting semantic understanding information. This information includes the degree of influence of the indicator on strategic goals, the correlation analysis with industry characteristics and business models, and the potential value and risk assessment of the indicator. For example, for the indicator "customer retention rate," the model can parse the following semantic understanding information: "This indicator is highly correlated with the corporate strategy's emphasis on 'improving customer satisfaction' and 'consolidating the existing customer base'; in the SaaS business model, customer retention rate directly affects long-term revenue and market share, and is one of the core indicators; in the cloud computing industry, due to intense competition, a high customer retention rate is crucial for building brand loyalty."

[0093] Furthermore, the target semantic reasoning model incorporates a weight allocation algorithm. This algorithm combines the semantic understanding information of all recommended indicators with key data from the enterprise profile (such as size and development stage) to perform comprehensive calculations and output a recommendation weight value for each indicator. The sum of the weight values ​​of all recommended indicators is 1, ensuring the standardization and rationality of the weight allocation. For example, after considering the above semantic understanding information, the model can output a recommendation weight value of 0.25 for the "customer retention rate" indicator, 0.20 for the "product innovation speed" indicator, and 0.15 for the "cost control effectiveness" indicator, and so on, until the sum of the weight values ​​of all indicators equals 1.

[0094] The steps outlined above enable the intelligent construction and personalized weight allocation of a corporate performance indicator system. By introducing advanced technologies such as knowledge graphs and large-scale language models (LLM), and combining them with corporate strategy, business characteristics, and industry trends, a performance indicator system that meets the company's actual needs and has a scientific basis, along with a corresponding weight allocation scheme, is automatically generated. This aims to improve the efficiency, accuracy, and scientific rigor of corporate performance management, enabling managers to more accurately assess corporate performance and guide strategic decision-making.

[0095] Step S204: Based on the recommended indicator set and indicator weight allocation scheme, generate the enterprise assessment scheme corresponding to the target enterprise.

[0096] It should be noted that the enterprise performance appraisal scheme corresponding to the target enterprise refers to a personalized and scientific performance appraisal scheme automatically generated based on the specific needs and strategic goals of the target enterprise, combined with the recommended set of performance appraisal indicators and the corresponding indicator weight allocation scheme. The scheme describes in detail the specific performance appraisal indicators, weight allocation, scoring rules and implementation details, aiming to provide clear guidance and basis for the enterprise's performance management.

[0097] In this embodiment of the invention, the recommended indicator set and the corresponding indicator weight allocation scheme are first integrated to ensure that each indicator has a clear weight value, reflecting its importance in the overall performance evaluation. Based on the integrated recommended indicators and weights, a clear evaluation framework is constructed, including defining the evaluation period (e.g., annual, quarterly), evaluation levels (enterprise level, department level, individual level), and scoring standards (e.g., target completion rate, peer comparison), making the evaluation scheme operable. Then, according to the evaluation framework, the specific components of the target company's evaluation scheme are generated: 1. Evaluation indicator description: Each evaluation indicator is defined in detail, including calculation formulas, scoring rules, and data sources. 2. Weight allocation description: The weight value of each evaluation indicator is clarified, and the basis and explanation for the weight allocation are provided, such as considerations based on corporate strategic goals, industry characteristics, or business models. 3. Implementation guidelines: The implementation steps, timelines, and responsible persons for the evaluation scheme are given to ensure that the scheme can be effectively implemented. 4. Adjustment mechanism: Dynamic adjustment rules for indicators and weights are set to adapt to changes in the corporate environment.

[0098] Furthermore, the steps for generating the enterprise assessment scheme corresponding to the target enterprise based on the recommended indicator set and indicator weight allocation scheme include: calling the preset assessment process framework, constructing the initial assessment scheme corresponding to the target enterprise based on the preset assessment process framework; and integrating each recommended indicator in the recommended indicator set and its corresponding recommended weight value into the initial assessment scheme to obtain the enterprise assessment scheme corresponding to the target enterprise.

[0099] It should be noted that the preset assessment process framework includes a general template for assessment schemes, including basic elements such as indicator classification, evaluation cycle, and scoring criteria. The system can retrieve assessment process frameworks from the stored template library that match the target company's size and industry characteristics. For example, for a growing technology company, the system will retrieve a framework suitable for innovation-driven companies that emphasizes quarterly target assessments.

[0100] Furthermore, based on the pre-defined assessment process framework, a basic, structured draft assessment plan is created. Information such as the classification of common indicators and hierarchical divisions within the framework can be filled into the plan template to form a preliminary, structured assessment plan framework. Simultaneously, some common evaluation cycles and data sources are pre-set according to the characteristics of the target company.

[0101] Furthermore, the recommended indicator set and its corresponding recommended weight values ​​are integrated into the initial assessment scheme to complete the personalized customization of the scheme. In specific implementation, each recommended indicator and its recommended weight value can be added to the corresponding indicator category to ensure that the weight allocation of each indicator reflects its importance in the overall performance evaluation. At the same time, detailed information such as the calculation logic and data source of each indicator is automatically filled in to form a complete corporate assessment scheme.

[0102] Alternatively, after generating the corporate performance evaluation plan, the system will provide an option that allows corporate managers to manually review and fine-tune the plan. Managers can adjust the indicators and weights based on the company's latest strategic goals and business needs to ensure the applicability and accuracy of the plan.

[0103] In addition to the intelligent screening and generation process based on large-scale language models and knowledge graphs described above, enterprises can also choose to generate assessment schemes using machine learning-based weight prediction models or expert systems. However, machine learning models may have an advantage in processing historical data, while expert systems excel in the interpretability of decisions. Nevertheless, these solutions may not achieve the level of intelligence achieved by the deep integration of artificial intelligence and expert knowledge described in this invention.

[0104] This invention enables intelligent generation and personalized matching of enterprise performance appraisal schemes, ensuring that the appraisal schemes not only reflect the enterprise's strategic goals but also accurately adapt to the enterprise's specific business scenarios and characteristics, thereby improving the efficiency and accuracy of performance management.

[0105] Through steps S201 to S204 above, the characteristic information of the target enterprise can be collected first, and an association model can be constructed based on the characteristic information and the preset assessment indicator list. The association model is used to store the relationship between characteristic items and indicator items. Then, based on the association model, N recommended assessment indicators are matched and selected from the full indicator library to form a recommended indicator set, where N is a positive integer. The recommended indicator set is then input into the target semantic reasoning model to output an indicator weight allocation scheme. In the indicator weight allocation scheme, each recommended assessment indicator corresponds to a recommended weight value. Finally, based on the recommended indicator set and the indicator weight allocation scheme, the enterprise assessment scheme corresponding to the target enterprise is generated.

[0106] This invention proposes a method for generating enterprise performance evaluation schemes. First, characteristic information of the target enterprise is collected. Then, based on the characteristic information and a pre-set list of performance evaluation indicators, an association model is constructed. This association model stores the relationships between characteristic items and indicator items. Next, based on the association model, N recommended performance evaluation indicators are matched and selected from the full indicator library to form a recommended indicator set, where N is a positive integer. The recommended indicator set is then input into the target semantic reasoning model, which outputs an indicator weight allocation scheme. In this scheme, each recommended performance evaluation indicator corresponds to a recommended weight value. Finally, based on the recommended indicator set and the indicator weight allocation scheme, an enterprise performance evaluation scheme corresponding to the target enterprise is generated.

[0107] The present invention will now be described in conjunction with another specific embodiment.

[0108] Figure 3 This is a module diagram of an optional intelligent recommendation system for enterprise performance evaluation schemes according to an embodiment of the present invention, such as... Figure 3 As shown, the system consists of an application interaction module, a data acquisition module, an indicator filtering and calculation module, and a data storage module.

[0109] 1. Application Interaction Module:

[0110] Receive enterprise identification information used to construct the assessment plan, obtain enterprise profile data corresponding to the selected enterprise based on the selected enterprise identification information, trigger the background indicator filtering and calculation engine based on the enterprise profile data, and return and display the personalized plan results in real time.

[0111] 2. Data Acquisition Module:

[0112] This module belongs to the data supply layer of the enterprise intelligent analysis system. It is specifically used to collect and aggregate the tag data and key indicator data from the full indicator library required to build enterprise profiles. This module includes the construction of enterprise profile data and the construction of the full indicator library, with the solutions as follows:

[0113] (1) Enterprise profiling construction scheme.

[0114] Multi-source external data is integrated, processed using Apache Flink, and used to construct enterprise profiles, which are then stored in a graph database (Neo4j is used as the graph database in this invention). Ultimately, a network structure that accurately reflects the connections between enterprises is formed and stored in the graph data.

[0115] (2) Construction plan for the full index database.

[0116] The construction of the full indicator library includes the acquisition and construction of specific indicator data, and its storage in a MongoDB database.

[0117] 3. Indicator screening and calculation module.

[0118] (1) Indicator Selection: The system collects discrete attributes (industry, stage, pain points), continuous attributes (scale, etc.), and textual attributes (strategic focus) of enterprises. The system uses a multi-dimensional fusion similarity algorithm to standardize and weightedly fuse these three types of attributes, selecting the top-N most similar enterprises as a reference set. Subsequently, the system uses frequency statistics to output high-frequency indicators for the indicators already used by enterprises in the reference set. Finally, the high-frequency indicator set is merged with indicators that match the characteristics of the enterprise and are directly inferred from the knowledge graph to form a recommended indicator set. This recommended indicator set is then used as context, along with the enterprise's strategic focus, compliance, and other information, and input into the LLM. The LLM performs a comprehensive evaluation based on this information and outputs a personalized indicator recommendation list and recommendation reasons, realizing intelligent and precise indicator selection in cold start scenarios. Finally, after manual adjustment, the final indicator set is obtained.

[0119] (2) Calculation of indicator weights: The initial weights of the indicators are calculated using the Analytic Hierarchy Process (AHP), and the weights are optimized and calibrated in combination with the large model. Then, they are manually adjusted to finally generate the assessment scheme.

[0120] 4. Data storage module.

[0121] The processed data is persisted, with the "enterprise profile" stored in the graph database Neo4j to store complex relationships, and the "full indicator library" stored in the MongoDB database to support efficient storage and querying of indicators.

[0122] This invention employs a combination of deep learning and graph analysis. By constructing precise corporate profiles and intelligently selecting assessment indicators, it achieves the goal of automatically generating scientific and personalized corporate assessment schemes. This improves the efficiency and accuracy of scheme formulation and solves the technical problem in related technologies where the corporate assessment scheme generation process lacks unified intelligent tools, resulting in low scheme formulation efficiency, reliance on subjective experience, and difficulty in coping with complex and ever-changing corporate environments and strategic needs.

[0123] Specifically, this invention first utilizes data acquisition and processing technologies to extract key feature information from multi-source data of the target enterprise, constructing a multimodal enterprise profile that comprehensively reflects the enterprise's current status and strategic orientation. Then, based on a highly structured association model, combined with the enterprise profile and a pre-set list of performance indicators, it quickly locates and filters a set of recommended performance indicators that highly match the enterprise's characteristics from the full indicator library. This process not only considers the direct correlation between indicators and enterprise characteristics but also utilizes historical data and the experience of similar enterprises to ensure the scientific and rational selection of indicators. Next, the selected set of recommended indicators is input into the target semantic reasoning model. Through deep semantic understanding and logical reasoning, an indicator weight allocation scheme is output. This scheme not only reflects the relative importance of indicators in enterprise performance evaluation but also considers the complex influence of enterprise strategy, industry characteristics, and business models, thereby avoiding subjectivity and arbitrariness in weight allocation and ensuring the objectivity and fairness of the evaluation scheme. Finally, based on the recommended indicator set and the indicator weight allocation scheme, a complete and accurate enterprise performance evaluation scheme is generated. This scheme not only accurately reflects the enterprise's strategic goals but also adapts to the dynamic changes in enterprise development, providing strong technical support for enterprise performance management and strategic execution. The above steps effectively improve the efficiency and quality of enterprise performance evaluation scheme development, realize the intelligentization of the entire process from scheme development to implementation, solve the problems of low efficiency, strong subjectivity and lack of scientific basis in the traditional scheme development process, and further solve the technical problem of low efficiency in scheme development caused by the lack of unified intelligent tools to support the enterprise performance evaluation scheme generation process.

[0124] The invention will now be described in conjunction with another alternative embodiment.

[0125] Example 2

[0126] The enterprise assessment scheme generation device provided in this embodiment includes multiple implementation units, each of which corresponds to a specific implementation step in Embodiment 1 above.

[0127] Figure 4 This is a schematic diagram of an optional enterprise performance evaluation scheme generation device according to an embodiment of the present invention, such as... Figure 4 As shown, the device may include: a data acquisition unit 41, a filtering unit 42, a reasoning unit 43, and a generation unit 44.

[0128] The acquisition unit 41 is used to acquire the characteristic information of the target enterprise and construct an association model based on the characteristic information and the preset assessment indicator list. The association model is used to store the relationship between the characteristic items and the indicator items.

[0129] The filtering unit 42 is used to match and filter N recommended assessment indicators from the full indicator library according to the association model to form a recommended indicator set, where N is a positive integer.

[0130] The reasoning unit 43 is used to input the set of recommendation indicators into the target semantic reasoning model and output the indicator weight allocation scheme, wherein each recommendation assessment indicator in the indicator weight allocation scheme corresponds to a recommendation weight value.

[0131] The generation unit 44 is used to generate the enterprise assessment scheme corresponding to the target enterprise based on the recommended indicator set and indicator weight allocation scheme.

[0132] The aforementioned enterprise assessment scheme generation device can collect the characteristic information of the target enterprise through the acquisition unit 41, and construct an association model based on the characteristic information and the preset assessment indicator list. The association model is used to store the relationship between characteristic items and indicator items. The filtering unit 42 matches and filters N recommended assessment indicators from the full indicator library according to the association model to form a recommended indicator set, where N is a positive integer. The inference unit 43 inputs the recommended indicator set into the target semantic inference model and outputs an indicator weight allocation scheme. In the indicator weight allocation scheme, each recommended assessment indicator corresponds to a recommended weight value. The generation unit 44 generates the enterprise assessment scheme corresponding to the target enterprise based on the recommended indicator set and the indicator weight allocation scheme.

[0133] In this embodiment of the invention, a method for generating an enterprise performance evaluation scheme is proposed. First, the characteristic information of the target enterprise is collected, and an association model is constructed based on the characteristic information and a preset list of performance evaluation indicators. The association model is used to store the association relationship between characteristic items and indicator items. Then, based on the association model, N recommended performance evaluation indicators are matched and selected from the full indicator library to form a recommended indicator set, where N is a positive integer. The recommended indicator set is then input into the target semantic reasoning model to output an indicator weight allocation scheme. In the indicator weight allocation scheme, each recommended performance evaluation indicator corresponds to a recommended weight value. Finally, based on the recommended indicator set and the indicator weight allocation scheme, the enterprise performance evaluation scheme corresponding to the target enterprise is generated.

[0134] This invention employs a combination of deep learning and graph analysis. By constructing precise corporate profiles and intelligently selecting assessment indicators, it achieves the goal of automatically generating scientific and personalized corporate assessment schemes. This improves the efficiency and accuracy of scheme formulation and solves the technical problem in related technologies where the corporate assessment scheme generation process lacks unified intelligent tools, resulting in low scheme formulation efficiency, reliance on subjective experience, and difficulty in coping with complex and ever-changing corporate environments and strategic needs.

[0135] Specifically, this invention first utilizes data acquisition and processing technologies to extract key feature information from multi-source data of the target enterprise, constructing a multimodal enterprise profile that comprehensively reflects the enterprise's current status and strategic orientation. Then, based on a highly structured association model, combined with the enterprise profile and a pre-set list of performance indicators, it quickly locates and filters a set of recommended performance indicators that highly match the enterprise's characteristics from the full indicator library. This process not only considers the direct correlation between indicators and enterprise characteristics but also utilizes historical data and the experience of similar enterprises to ensure the scientific and rational selection of indicators. Next, the selected set of recommended indicators is input into the target semantic reasoning model. Through deep semantic understanding and logical reasoning, an indicator weight allocation scheme is output. This scheme not only reflects the relative importance of indicators in enterprise performance evaluation but also considers the complex influence of enterprise strategy, industry characteristics, and business models, thereby avoiding subjectivity and arbitrariness in weight allocation and ensuring the objectivity and fairness of the evaluation scheme. Finally, based on the recommended indicator set and the indicator weight allocation scheme, a complete and accurate enterprise performance evaluation scheme is generated. This scheme not only accurately reflects the enterprise's strategic goals but also adapts to the dynamic changes in enterprise development, providing strong technical support for enterprise performance management and strategic execution. The above steps effectively improve the efficiency and quality of enterprise performance evaluation scheme development, realize the intelligentization of the entire process from scheme development to implementation, solve the problems of low efficiency, strong subjectivity and lack of scientific basis in the traditional scheme development process, and further solve the technical problem of low efficiency in scheme development caused by the lack of unified intelligent tools to support the enterprise performance evaluation scheme generation process.

[0136] Furthermore, the acquisition unit includes: a receiving module for receiving enterprise identification information transmitted by the user through an interactive interface, wherein the enterprise identification information is used to identify the target enterprise; a first extraction module for periodically calling the data acquisition engine to obtain data stream information, and extracting the feature information of the target enterprise from the data stream information based on the enterprise identification information; a first construction module for constructing an enterprise profile of the target enterprise based on the feature information, wherein the enterprise profile is presented in the form of multimodal features, including: category attributes, numerical attributes, and text attributes; and a second construction module for constructing a correlation model corresponding to the target enterprise based on a preset list of assessment indicators and the enterprise profile.

[0137] Furthermore, the second construction module includes: a first construction submodule, used to construct enterprise feature nodes and assign values ​​to enterprise feature nodes based on multimodal feature items in the enterprise profile to obtain first node information; a second construction submodule, used to construct assessment indicator nodes and assign values ​​to assessment indicator nodes based on preset assessment indicators in a preset assessment indicator list to obtain second node information; a first calculation submodule, used to calculate the similarity between the first node information and the second node information for each enterprise feature node and each assessment indicator node using a similarity algorithm, and determine the similarity as an association weight value, wherein the magnitude of the association weight value is used to characterize the strength of the association between feature items and indicator items; and an establishment submodule, used to establish the association relationship between each enterprise feature node and each assessment indicator node based on the association weight value to obtain the association model corresponding to the target enterprise.

[0138] Furthermore, the enterprise performance evaluation scheme generation device also includes: a construction unit for constructing a full-scale indicator library, the construction unit comprising: a second extraction module for extracting historical performance evaluation schemes from the system database, wherein the historical performance evaluation schemes cover multiple different enterprises, different development stages, and different business models; an analysis module for analyzing, for each historical performance evaluation scheme, the historical enterprise entities in the historical performance evaluation scheme, the historical characteristic information possessed by the historical enterprise entities, the historical performance evaluation indicators that match the historical characteristic information in the historical performance evaluation scheme, and all historical performance evaluation indicators used in the historical performance evaluation scheme; an establishment module for establishing a first mapping relationship between historical enterprise entities and historical characteristic information, a second mapping relationship between historical characteristic information and historical performance indicators, a third mapping relationship between historical performance evaluation schemes and historical performance indicators, and a fourth mapping relationship between historical enterprise entities and historical performance evaluation schemes; and a third construction module for constructing a full-scale indicator library based on the first, second, third, and fourth mapping relationships.

[0139] Furthermore, the screening unit includes: a first determining module, used to determine historical feature information matching the target enterprise from the full indicator library by combining the association relationship indicated by the association model and the first mapping relationship in the full indicator library; a screening module, used to retrieve the enterprise profile of the target enterprise, and screen the historical feature information based on the multimodal features in the enterprise profile to obtain the first recommended feature set of the target enterprise; a second determining module, used to determine the historical assessment indicators corresponding to the recommended features from the full indicator library based on the recommended feature set and the second mapping relationship to obtain the first type of recommended indicators; a third determining module, used to determine the historical assessment scheme corresponding to the historical enterprise entity based on the fourth mapping relationship when the full indicator library contains historical enterprise entities matching the target enterprise, and to determine the historical assessment indicators involved in the historical assessment scheme based on the third mapping relationship to obtain the second type of recommended indicators; and a first integrating module, used to integrate the first type of recommended indicators and the second type of recommended indicators to obtain a recommended indicator set containing N recommended assessment indicators.

[0140] Furthermore, the filtering module includes: an extraction submodule, used to extract multimodal features to obtain categorical attribute features, numerical attribute features, and textual attribute features, and to classify historical feature information into categorical attribute historical features, numerical attribute historical features, and textual attribute historical features; a second calculation submodule, used to calculate the cross-matching degree between the categorical attribute features of the target company and each categorical attribute historical feature using a cross-matching algorithm, to obtain a first matching degree value; a third calculation submodule, used to calculate the cosine similarity between the numerical attribute features of the target company and each numerical attribute historical feature using a cosine similarity algorithm, to obtain a second matching degree value; a fourth calculation submodule, used to map the textual attribute features of the target company to textual feature vectors and the textual attribute historical features to textual historical feature vectors using a vectorization algorithm, and to calculate the cosine similarity between the textual feature vectors and each textual historical feature vector using a cosine similarity algorithm, to obtain a third matching degree value; and a filtering submodule, used to filter historical feature information corresponding to matching degree values ​​greater than or equal to a preset value, to obtain a first recommended feature set for the target company.

[0141] Furthermore, the reasoning unit includes: a semantic understanding module, used to call the auxiliary reasoning text of the target enterprise using the target semantic reasoning model, and to perform semantic understanding on each recommendation indicator in the recommendation indicator set based on the auxiliary reasoning text to obtain semantic understanding information. The auxiliary reasoning text is pre-stored in the database of the target semantic reasoning model and includes the target enterprise's corporate strategy information, industry characteristic information, and business model information. The semantic understanding information at least includes the impact of the recommendation indicators on the corporate strategy, industry characteristics, and business model. A calculation module is used to calculate the recommendation weight value corresponding to each recommendation indicator by using the weight allocation algorithm pre-built in the target semantic reasoning model and combining the semantic understanding information corresponding to all recommendation indicators, to obtain the indicator weight allocation scheme. The sum of the recommendation weight values ​​corresponding to all recommendation indicators is 1.

[0142] Furthermore, the generation unit includes: a fourth construction module, used to call a preset assessment process framework and construct an initial assessment scheme for the target enterprise based on the preset assessment process framework; and a second integration module, used to integrate each recommended indicator in the recommended indicator set and its corresponding recommended weight value into the initial assessment scheme to obtain the enterprise assessment scheme for the target enterprise.

[0143] The aforementioned enterprise assessment scheme generation device may also include a processor and a memory. The aforementioned acquisition unit 41, screening unit 42, reasoning unit 43, generation unit 44, etc., are all stored in the memory as program units, and the processor executes the aforementioned program units stored in the memory to realize the corresponding functions.

[0144] The aforementioned processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured. By adjusting kernel parameters, characteristic information of the target company is collected. Based on this characteristic information and a pre-defined list of performance indicators, a correlation model is constructed. The correlation model then matches and filters recommended performance indicators from the full indicator library, forming a recommended indicator set. This recommended indicator set is input into the target semantic reasoning model, which outputs an indicator weight allocation scheme. Based on the recommended indicator set and the indicator weight allocation scheme, a corresponding performance evaluation scheme for the target company is generated.

[0145] The aforementioned memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0146] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialization program with the following method steps: collecting feature information of the target enterprise, and constructing an association model based on the feature information and a preset list of assessment indicators, wherein the association model is used to store the association relationship between feature items and indicator items; matching and filtering N recommended assessment indicators from the full indicator library according to the association model to form a recommended indicator set, wherein N is a positive integer; inputting the recommended indicator set into the target semantic reasoning model and outputting an indicator weight allocation scheme, wherein each recommended assessment indicator in the indicator weight allocation scheme corresponds to a recommended weight value; and generating an enterprise assessment scheme corresponding to the target enterprise based on the recommended indicator set and the indicator weight allocation scheme.

[0147] According to another aspect of the present invention, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored computer program, wherein, when the computer program is running, it controls the device where the computer-readable storage medium is located to execute the method for generating an enterprise assessment scheme according to any one of the above embodiments.

[0148] According to another aspect of the present invention, an electronic device is also provided, including one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the enterprise assessment scheme generation method of any one of the above embodiments.

[0149] Figure 5 This is a structural block diagram of an electronic device for executing a method for generating an enterprise performance evaluation scheme according to an embodiment of the present invention, such as... Figure 5 As shown, the electronic device may include: one or more ( Figure 5 (Only one is shown) processor 502, memory 504, memory controller, and peripheral interface, wherein the peripheral interface is connected to the radio frequency module, audio module and display.

[0150] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the enterprise assessment scheme generation method and apparatus in this application embodiment. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby realizing the aforementioned enterprise assessment scheme generation method. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0151] Those skilled in the art will understand that Figure 5 The structure shown is for illustrative purposes only. Electronic devices can also be smartphones, tablets, handheld computers, mobile internet devices (MIDs), PADs, and other terminal devices. Figure 5 This does not limit the structure of the aforementioned electronic device. For example, electronic devices may also include components that are more... Figure 5 The more or fewer components shown (such as network interfaces, display devices, etc.), or having the same Figure 5 The different configurations shown.

[0152] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0153] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0154] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0155] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0156] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0157] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0158] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0159] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for generating an enterprise performance evaluation scheme, characterized in that, include: Collect characteristic information of the target enterprise, and construct an association model based on the characteristic information and a preset list of assessment indicators. The association model is used to store the association relationship between characteristic items and indicator items. Based on the aforementioned association model, N recommended assessment indicators are matched and selected from the full indicator library to form a recommended indicator set, where N is a positive integer; The set of recommended indicators is input into the target semantic reasoning model, and an indicator weight allocation scheme is output, wherein each of the recommended assessment indicators in the indicator weight allocation scheme corresponds to a recommended weight value; Based on the recommended indicator set and the indicator weight allocation scheme, a corporate performance evaluation scheme corresponding to the target enterprise is generated.

2. The generation method according to claim 1, characterized in that, The steps of collecting characteristic information of target enterprises and constructing an association model based on the characteristic information and a preset list of performance indicators include: Receive enterprise identification information transmitted by the user through an interactive interface, wherein the enterprise identification information is used to identify the target enterprise; The data acquisition engine is periodically invoked to obtain data stream information, and the characteristic information of the target enterprise is extracted from the data stream information based on the enterprise identification information. The enterprise profile of the target enterprise is constructed based on the feature information, wherein the enterprise profile is presented in the form of multimodal features, and the multimodality includes: category attributes, numerical attributes and text attributes; Based on the preset assessment indicator list and the enterprise profile, the association model corresponding to the target enterprise is constructed.

3. The generation method according to claim 2, characterized in that, The steps for constructing the association model corresponding to the target enterprise based on the preset assessment indicator list and the enterprise profile include: Construct enterprise feature nodes, and assign values ​​to the enterprise feature nodes based on the multimodal feature items in the enterprise profile to obtain the first node information; Construct assessment indicator nodes, and assign values ​​to the assessment indicator nodes based on the preset assessment indicators in the preset assessment indicator list to obtain the second node information; For each of the enterprise feature nodes and each of the assessment indicator nodes, a similarity algorithm is used to calculate the similarity between the first node information and the second node information, and the similarity is determined as the association weight value, wherein the magnitude of the association weight value is used to characterize the strength of the association between the feature item and the indicator item; Based on the association weight values, establish the association relationship between each enterprise feature node and each assessment indicator node to obtain the association model corresponding to the target enterprise.

4. The generation method according to claim 1, characterized in that, The complete indicator library is constructed through the following steps: Historical assessment schemes are extracted from the system database, which cover multiple different enterprises, different stages of development, and different business models. For each historical assessment scheme, analyze the historical enterprise entity in the historical assessment scheme, the historical characteristic information of the historical enterprise entity, the historical assessment indicators that match the historical characteristic information in the historical assessment scheme, and all the historical assessment indicators used in the historical assessment scheme. A first mapping relationship is established between the historical enterprise entity and the historical feature information; a second mapping relationship is established between the historical feature information and the historical performance indicators; a third mapping relationship is established between the historical performance evaluation scheme and the historical performance indicators; and a fourth mapping relationship is established between the historical enterprise entity and the historical performance evaluation scheme. Based on the first mapping relationship, the second mapping relationship, the third mapping relationship, and the fourth mapping relationship, the full index library is constructed.

5. The generation method according to claim 4, characterized in that, The steps for matching and selecting N recommended assessment indicators from the full indicator library to form a recommended indicator set based on the aforementioned association model include: By combining the association relationship indicated by the association model and the first mapping relationship in the full index library, historical feature information matching the target enterprise is determined from the full index library; Retrieve the enterprise profile of the target enterprise, and filter the historical feature information based on the multimodal features in the enterprise profile to obtain the first recommended feature set of the target enterprise; Based on the recommended feature set and the second mapping relationship, the historical assessment indicators corresponding to the recommended features are determined from the full indicator library to obtain the first type of recommended indicators; If the historical enterprise entity that matches the target enterprise is included in the full index library, the historical assessment scheme corresponding to the historical enterprise entity is determined based on the fourth mapping relationship, and the historical assessment indicators involved in the historical assessment scheme are determined based on the third mapping relationship to obtain the second type of recommended indicators. By integrating the first type of recommendation indicators and the second type of recommendation indicators, a set of recommendation indicators containing N of the aforementioned recommendation assessment indicators is obtained.

6. The generation method according to claim 5, characterized in that, The step of filtering the historical feature information based on the multimodal features in the enterprise profile to obtain the first recommended feature set of the target enterprise includes: Extract the multimodal features to obtain category attribute features, numerical attribute features and text attribute features, and classify the historical feature information into category attribute historical features, numerical attribute historical features and text attribute historical features; The cross-matching algorithm is used to calculate the cross-matching degree between the category attribute features of the target enterprise and the historical features of each category attribute, and a first matching degree value is obtained; The cosine similarity algorithm is used to calculate the cosine similarity between the numerical attribute features of the target enterprise and the historical features of each numerical attribute, and a second matching degree value is obtained. Using a vectorization algorithm, the text attribute features of the target enterprise are mapped to text feature vectors, and the text attribute history features are mapped to text history feature vectors. The cosine similarity algorithm is then used to calculate the cosine similarity between the text feature vector and each text history feature vector to obtain a third matching degree value. By filtering the historical feature information corresponding to matching degree values ​​greater than or equal to a preset value, the first recommended feature set of the target enterprise is obtained.

7. The generation method according to claim 1, characterized in that, The steps of inputting the recommended indicator set into the target semantic reasoning model and outputting the indicator weight allocation scheme include: The target semantic reasoning model is used to call the auxiliary reasoning text of the target enterprise, and semantic understanding is performed on each recommendation indicator in the recommendation indicator set based on the auxiliary reasoning text to obtain semantic understanding information. The auxiliary reasoning text is pre-stored in the database of the target semantic reasoning model. The auxiliary reasoning text includes the enterprise strategy information, industry characteristic information and business model information of the target enterprise. The semantic understanding information includes at least the impact of the recommendation indicators on the enterprise strategy, industry characteristics and business model. Using a weight allocation algorithm pre-installed in the target semantic reasoning model, and combining the semantic understanding information corresponding to all the recommendation indicators, the recommendation weight value corresponding to each recommendation indicator is calculated to obtain the indicator weight allocation scheme, wherein the sum of the recommendation weight values ​​corresponding to all the recommendation indicators is 1.

8. The generation method according to claim 1, characterized in that, The steps for generating a corporate performance evaluation scheme for the target company based on the recommended indicator set and the indicator weight allocation scheme include: Invoke the preset assessment process framework, and construct the initial assessment plan corresponding to the target enterprise based on the preset assessment process framework; Each of the recommended indicators in the recommended indicator set and its corresponding recommended weight value are integrated into the initial assessment scheme to obtain the enterprise assessment scheme corresponding to the target enterprise.

9. A device for generating an enterprise performance evaluation scheme, characterized in that, include: The data collection unit is used to collect the characteristic information of the target enterprise and construct an association model based on the characteristic information and a preset list of assessment indicators. The association model is used to store the association relationship between the characteristic items and the indicator items. The filtering unit is used to match and filter N recommended assessment indicators from the full indicator library according to the association model to form a recommended indicator set, where N is a positive integer; The inference unit is used to input the set of recommendation indicators into the target semantic inference model and output an indicator weight allocation scheme, wherein each of the recommendation assessment indicators in the indicator weight allocation scheme corresponds to a recommendation weight value; The generation unit is used to generate the enterprise assessment scheme corresponding to the target enterprise based on the recommended indicator set and the indicator weight allocation scheme.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the method for generating an enterprise assessment scheme as described in any one of claims 1 to 8.

Images