Enterprise digital diagnosis method and system based on process-object double drive and knowledge graph

Abstract

The invention provides an enterprise digital diagnosis method and system based on process-object double drive and a knowledge graph. The method comprises the following steps: constructing an industry knowledge graph; taking the process as a basic evaluation item and the object as a corresponding evaluation index to form a process-object double-drive evaluation system; acquiring digital transformation information of a target enterprise, and matching a knowledge graph and an evaluation system for the digital transformation information; existing object data are extracted from the digital transformation information; digitization levels are divided based on the relation between the existing object data and the object set, an overall score is obtained based on the relation between the existing object data and the industry reference value in the knowledge graph, and finally the comprehensive maturity level is obtained. By setting the knowledge graph and the evaluation system based on the process-object double drive, quantitative analysis of the digital coverage rate of the whole process of the enterprise and digital deep evaluation of the multi-level object are realized, and the digital weak link of the enterprise can be accurately positioned.

Description

Technical Field

[0001] The present application relates to the technical field of digital transformation, and particularly relates to an enterprise digital diagnosis method and system based on process-object dual drive and knowledge graph. Background Art

[0002] Digital transformation refers to the complete or partial transformation of traditional business processes, organizational structures, marketing methods, etc. into digital forms to improve the efficiency, innovation ability, and competitiveness of enterprises. At present, digital transformation has become an inevitable trend in enterprise development.

[0003] The current digital maturity assessment of enterprises mainly relies on expert experience and questionnaires, and macro-judgments on the digital level of enterprises are made through the fuzzy evaluation method. This method has the following disadvantages: 1) Excessive subjectivity: The proportion of expert experience is high, and it is easily affected by individual differences, resulting in inaccurate evaluation results; 2) Low efficiency: It relies on manual questionnaire filling, has a long diagnosis cycle, and cannot provide real-time feedback; 3) Lack of dynamic association: It does not combine the industrial chain map and knowledge graph, and cannot analyze the digital level from the perspectives of the whole process and all objects.

[0004] It should be noted that the above introduction to the technical background is only for the convenience of clearly and completely explaining the technical solutions of the present application and facilitating the understanding of those skilled in the art. It cannot be considered that the above technical solutions are well-known to those skilled in the art just because these solutions are described in the background art part of the present application. Summary of the Invention

[0005] To solve the above problems, on the one hand, the present application provides an enterprise digital diagnosis method based on process-object dual drive and knowledge graph, including:

[0006] Constructing an industry knowledge graph based on industry knowledge, where the industry knowledge graph includes several sub-industry knowledge graphs;

[0007] For each sub-industry, subdivide the business process of the enterprise into multiple processes, and set an object set for each process, where the object set includes at least one object, and form an evaluation system based on process-object dual drive with the process as the basic evaluation item and the object as the corresponding evaluation index;

[0008] Collecting the digital transformation information of the target enterprise, and matching the corresponding sub-industry knowledge graph and evaluation system for the target enterprise;

[0009] Based on the preset processes and object sets in the matched evaluation system, extracting the object data existing in the target enterprise for each process from the digital transformation information;

[0010] Divide the digital levels of the target enterprise according to preset rules based on the relationship between the existing object data in the target enterprise and the object set;

[0011] Obtain an overall score based on the relationship between the existing object data in the target enterprise and the industry benchmark values in the industry knowledge graph of the sub - industry;

[0012] Obtain the comprehensive maturity level of the target enterprise based on the digital level and the overall score;

[0013] This application realizes the quantitative analysis of the digital coverage rate of the entire enterprise process and the digital depth assessment of multi - level objects by setting up a knowledge graph and an evaluation system driven by process - object dual - drive, improves the matching degree of the evaluation results with the actual enterprise scenarios, and can accurately locate the weak links in enterprise digitization.

[0014] The construction process of the industry knowledge graph includes: obtaining data from the industry knowledge base; cleaning the data obtained from the industry knowledge base; extracting <entity, relationship, attribute> triples from the cleaned data; constructing a domain ontology covering five major links of design, production, supply chain, sales, and service.

[0015] The industry knowledge graph dynamically associates the digital levels of enterprises in the sub - industry in five links of design, production, supply chain, sales, and service, and deploys a graph neural network for relationship reasoning.

[0016] The industry knowledge graph will be iteratively updated based on the newly added enterprise data, including: constructing a LoRA fine - tuning layer, injecting the enterprise data into the LLaMA - 2 model; visually identifying key features through the attention mechanism, and automatically optimizing the node weights of the industry knowledge graph.

[0017] The digital transformation information includes basic information and business process data; the basic information includes at least the industry type, scale, and main business scenarios of the target enterprise; the business process data includes equipment - layer data, edge - layer data, and enterprise - layer data.

[0018] Match the corresponding industry knowledge graph and evaluation system of the target enterprise based on the basic information; the matching of the industry knowledge graph of the sub - industry is realized by using the NLP matching algorithm.

[0019] Calculate the overall digital degree of the processes of the target enterprise based on the ratio of the number of existing object data in the target enterprise to the number of objects included in the object set, and divide the digital levels of the target enterprise based on the overall digital degree of the processes and preset rules; the preset rules include setting five - level digital levels and corresponding scoring intervals, and determining the digital levels of the target enterprise according to the overall digital degree of the processes and the scoring intervals.

[0020] The industry benchmark value is calculated based on the data of the top 10% of the enterprises in the industry; a quantitative score is obtained based on the deviation between the existing object data in the target enterprise and the industry benchmark value to obtain an overall score.

[0021] Based on the digitalization level and the overall score, a weighted calculation is performed using the random forest algorithm to obtain the comprehensive maturity level of the target enterprise.

[0022] On the other hand, the present application also provides an enterprise digital diagnosis system based on process-object dual drive and knowledge graph, including: the system includes a data collection module, an industry knowledge graph module, and a dual drive evaluation module; the industry knowledge graph module includes an industry knowledge graph constructed based on industry knowledge, and the industry knowledge graph includes several sub-industry knowledge graphs; the dual drive evaluation module includes an evaluation system based on process-object dual drive for each sub-industry, and the evaluation system based on process-object dual drive includes multiple processes obtained by subdividing the business processes of the enterprise and an object set corresponding to each process, and the object set includes at least one object; the data collection module is used to collect the digital transformation information of the target enterprise and extract the existing object data of each process in the target enterprise; the dual drive evaluation module matches a corresponding evaluation system for the target enterprise based on the digital transformation information, and determines the digitalization level of the target enterprise based on the relationship between the existing object data in the target enterprise and the object set; the industry knowledge graph module matches a corresponding sub-industry knowledge graph for the target enterprise based on the digital transformation information, and determines the overall score based on the relationship between the existing object data in the target enterprise and the industry benchmark value in the sub-industry knowledge graph; it further includes a diagnosis report generation module for generating a diagnosis report according to the processing results of the industry knowledge graph module and the dual drive evaluation module.

[0023] Compared with the prior art, the beneficial effects of the present invention mainly include the following: 1) Precise diagnosis: Through the quantification of process coverage rate and the in-depth scoring of objects, it determines the presence or absence by process and evaluates the degree by object, reducing the diagnosis error by 30%; 2) Dynamic adaptation: The knowledge graph is automatically updated with industrial data, shortening the model iteration cycle by 50%; 3) Efficiency improvement: The diagnosis time is compressed from the traditional 7 days to 2 hours. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the specific embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following described drawings are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other drawings without creative efforts based on these drawings.

[0025] Figure 1 Schematic diagram of the diagnostic method provided by the present invention. Detailed implementation manners

[0026] Regarding the foregoing and other technical contents, features and effects of the present invention, they will be clearly presented in the following detailed description of a preferred embodiment with reference to the drawings. Directional terms mentioned in the following embodiments, such as: up, down, left, right, front or back, etc., are only with reference to the directions of the attached drawings. Therefore, the directional terms used are for illustration and not for limiting the present invention.

[0027] The following will elaborate on each embodiment of the present application with reference to the drawings. However, those of ordinary skill in the art can understand that in each embodiment of the present application, many technical details are proposed for the reader to better understand the present application. However, even without these technical details and various changes and modifications based on the following embodiments, the technical solutions claimed in the present application can still be implemented.

[0028] The steps in the following embodiments do not correspond one by one to the content of the invention.

[0029] Embodiment 1

[0030] Figure 1 Schematic diagram of the diagnostic method provided by the present invention. Refer to Figure 1 , the present invention provides an enterprise digital diagnosis method based on process-object dual drive and knowledge graph. The method includes the following steps:

[0031] Step 1: Construct an industry knowledge graph based on industry knowledge.

[0032] Existing methods for evaluating the digital maturity of enterprises mostly rely on expert experience and questionnaires. The index scoring is highly subjective, and the diagnostic results lack sufficient objectivity and fairness. On the other hand, for different industries, different evaluation criteria should be adopted. To improve the objectivity of diagnosis and its adaptability to different industries, this application proposes to diagnose based on the current situation and industry standards of the same industry of the enterprise to be diagnosed. To make full use of the existing data in the industry, the knowledge scattered in the industry can be constructed into a knowledge graph. Through in-depth semantic analysis and data mining, the knowledge graph can efficiently organize a large amount of industrial Internet data into a knowledge network, search and display knowledge in an intuitive way, and at the same time provide important guarantees for big data analysis, intelligent judgment, personalized recommendation, etc.

[0033] The construction of an industry knowledge graph generally includes processes such as data collection, data preprocessing, knowledge extraction, ontology construction, knowledge fusion, and knowledge storage. It can be understood that based on the differences in industries, their evaluation criteria should also be different. Therefore, a corresponding knowledge graph should be generated for each sub-industry, that is, it can be considered that a corresponding sub-industry knowledge graph is constructed for different sub-industries, and the sum of all sub-industry knowledge graphs is the industry knowledge graph. The following introduces the construction of the industry knowledge graph.

[0034] First is data collection, which is the starting step in constructing an industry knowledge graph and involves obtaining data from various sources. Data sources can be divided into structured data (such as databases), semi-structured data (such as HTML pages, XML files), and unstructured data (such as text, images, etc.). It can be understood that data collection can be carried out by oneself for collection and collation, or obtained from some existing databases. An industry knowledge base is a system used by enterprises or within a specific industry to store, manage, and share professional knowledge, experience, and processes. In this application, data is obtained from industry knowledge bases (such as knowledge bases for mold manufacturing process flows, chemical reaction chains, etc.).

[0035] Then comes data preprocessing, which usually includes steps such as data cleaning and entity recognition, with the aim of converting the original data into a format suitable for constructing a knowledge graph. Data cleaning usually involves removing errors, duplicates, or incomplete information, such as noise removal, data normalization, and missing value handling. In this application, the BERT-Whitening algorithm was used to clean the data in the industry knowledge base.

[0036] Next is knowledge extraction, which extracts <entity, relationship, attribute> triples from the preprocessed data. Common methods include: entity extraction: identifying entities in the text, which can use rule-based methods, machine learning methods, or deep learning methods; relationship extraction: identifying the relationships between entities, and methods include rule-based extraction, machine learning methods, and deep learning methods; attribute extraction: extracting the attribute information of entities.

[0037] Then comes ontology construction. An ontology is a set of terms and definitions used in a knowledge graph to describe the knowledge and concepts in a specific domain, defining the entity categories, attributes, and relationship types in the knowledge graph. In this application, the ontology editing tool Protégé was used to construct a domain ontology covering 5 major links of design, production, supply chain, sales, and service.

[0038] Subsequent steps also include knowledge fusion and knowledge storage. Knowledge fusion refers to integrating knowledge from multiple sources to eliminate contradictions and ambiguities; while knowledge storage is to store the extracted and fused knowledge into a graph database. In this application, the Neo4j graph database was adopted, which is a high-performance NoSQL graph database suitable for processing complex relational data.

[0039] The above is the construction process of the industry knowledge graph, and the industry knowledge graph at this time can be used for subsequent digital diagnosis processes.

[0040] Furthermore, after constructing the industry knowledge graph, it can be dynamically associated and optimized. On the one hand, in this embodiment, by dynamically associating the digital levels of enterprises in each industry in 5 links such as design, production, supply chain, sales, and service, a graph neural network (GNN) is deployed for relationship reasoning. Generally, in graph-structured data, nodes represent entities and edges represent relationships between entities. GNN can effectively perform relationship reasoning by learning the representations of nodes and edges. For example, taking the supply chain link as an example, the following rules can be set: when it is detected that the work order completion rate of the MES system > 95%, automatically enhance the connection weight with the supply chain node; further, use a reinforcement learning framework (PPO algorithm) to optimize the node priority, and the reward function is set as Δ (process efficiency gain / resource consumption).

[0041] In this application, a graph database (Neo4j) is used to manage the industry knowledge graph. Neo4j is used to store and update the relationships between nodes (such as entities like devices and processes) and edges (such as relationships like data flow and dependency relationships), and can support time series attributes (such as device status change records).

[0042] On the other hand, it can be understood that the construction of the knowledge graph is an iterative process and can be continuously iteratively updated according to new data and user feedback. In this application, by analyzing the newly added enterprise data through a large model (Transformer), on the one hand, a LoRA fine-tuning layer is constructed to inject enterprise real-time data (such as device logs, work order records, etc.) into the LLaMA-2 model; on the other hand, key features are identified through attention mechanism visualization (Transformer-Explain tool) to automatically optimize the graph node weights (for example, after the digitalization rate of a certain link increases, the priority of its associated nodes is adjusted), including indicators such as node connection entropy and relationship density.

[0043] Step 2: Set up an evaluation system driven by both process and object.

[0044] The diagnostic method provided in this application is carried out based on two dimensions: process and object. In this application, the process refers to the basic evaluation items obtained after multi-level subdivision of the production and operation activities of an enterprise (i.e., business processes), and the object refers to the evaluation indicators corresponding to each basic evaluation item. That is, the business processes of an enterprise are subdivided into multiple processes, and several objects are correspondingly set for each process, thus forming an evaluation system driven by both process and object. It can be understood that for enterprises in different industries, the specific content included in the above evaluation system may be different. For enterprises within the same industry, an evaluation system with the same standard can be adopted.

[0045] Specifically, the business processes of an enterprise are first divided into five links: design, production, supply chain, sales, and service; each link is further divided into several types; each type is divided into several scenarios; and finally each scenario is divided into several processes.

[0046] In this application, the object is an evaluation indicator established to judge the digitalization degree of each process, and the establishment of this evaluation indicator is based on the summary of experience in actual production and operation activities. Each process may have multiple objects, and the set of all objects corresponding to a process is called an object set, that is, each process is provided with a corresponding object set. It can be understood that for a process, if the process is digitalized, some corresponding data will inevitably be generated. For an enterprise, this data can be divided into three categories, which respectively exist in device layer data, edge layer data, and enterprise layer data. Therefore, the existing objects can be extracted from the device layer data, edge layer data, and enterprise layer data to judge the digitalization degree of this process.

[0047] Step 3: Collect the digital transformation information of the target enterprise and match the corresponding industry knowledge graph and evaluation system of the target enterprise.

[0048] In this application, the enterprise to be diagnosed for digital maturity is called the target enterprise, and the data required for digital maturity diagnosis is called digital transformation information. It can be understood that the digital transformation information of the target enterprise can be provided by the target enterprise; further, to improve the accuracy, the target enterprise can also be allowed to provide relevant data interfaces of the enterprise for automatic collection. The digital transformation information may include: basic information (such as the industry type, scale, main business scenarios, etc. of the target enterprise) and business process data, etc.

[0049] In this application, enterprise information can be entered first. For example, the target enterprise can be asked to fill in basic information such as industry type, scale, and main business scenarios. Then, according to the basic information, the corresponding sub-industry knowledge graph and evaluation system of the sub-industry to which the target enterprise belongs can be automatically associated and matched. In this application, the matching of the knowledge graph is implemented using the NLP matching algorithm, which can automatically recommend the most relevant sub-industry knowledge graph.

[0050] In this application, multi-source data access to the target enterprise can also be performed to further collect business process data. According to the source of the business process data, it can be divided into three levels of data. 1) Device layer data: The operating status of devices and sensor data are collected through industrial Internet of Things protocols (such as OPC UA or MQTT protocols). Through this part of the data, information such as the machine tool networking rate and sensor coverage rate of the target enterprise can be obtained. A feasible approach is to use an OPC UA proxy server to perform lossless protocol conversion on numerically controlled machine tools, with a sampling frequency ≥ 100Hz; 2) Edge layer data: Apache Kafka edge nodes are deployed to implement data stream window function calculations and obtain the real-time processing results of edge computing nodes; 3) Enterprise layer data: Structured data such as the digitalization rate of business documents and process coverage rate are extracted by docking enterprise management systems such as ERP and MES through APIs. A feasible approach is to build an adapter matrix through the OpenAPI 3.0 standard, which can support field-level mapping of ERP systems such as SAP / UFIDA / Kingdee.

[0051] Step 4: Extract object data from the digital transformation data, and combine it with the matched sub-industry knowledge graph and the process-object dual-driven evaluation system to conduct a digital maturity diagnosis.

[0052] It can be understood that if a process is digitized, some corresponding data will inevitably be generated during the implementation of the process, that is, the object data of the process exists in the target enterprise. For an enterprise, object data can be divided into three categories according to the source, corresponding to those existing in the device layer data, edge layer data, and enterprise layer data respectively. Therefore, the object data that already exists in the target enterprise can be extracted from the digital transformation data (including the collected device layer data, edge layer data, and enterprise layer data), and by comparing the object data of each process that already exists in the target enterprise with the object set pre-established in the evaluation system, the digitalization degree of the target enterprise's processes can be evaluated.

[0053] Specifically, according to the preset processes and corresponding object sets in the process-object dual-driven evaluation system, the object data that already exists in the target enterprise for each process is extracted from the previously obtained digital transformation information of the target enterprise. Then, the object data that already exists in the target enterprise for each process is compared with the object set of that process. The larger the number of the extracted existing object data, the higher the digitalization level of that process. A feasible calculation method is: Process digitalization level = the number of object data that already exists in the target enterprise for that process / the number of objects included in the object set of that process.

[0054] In fact, in this application, there is no limitation on how to extract the existing object data from the digital transformation information. For example, the method of manual confirmation can be adopted, or it can be automatically obtained with the help of an algorithm, or it can be obtained by combining manual work and an algorithm. A feasible method is: Let the target enterprise check the existing objects for each process through a visual interface; further, in order to improve objectivity and accuracy, the existing object data can also be automatically obtained with the help of an algorithm.

[0055] On the one hand, the digital level of the target enterprise can be obtained based on the existing object data in the target enterprise. Specifically, the digitalization level of each process can be obtained by comparing the extracted existing object data with the object set; further, based on the digitalization level of each process, the digitalization level of each scenario can be obtained; based on the digitalization level of each scenario, the digitalization level of each type can be obtained; based on the digitalization level of each type, the digitalization level of each link can be obtained; based on the digitalization level of each link, the overall process digitalization level of the target enterprise can be obtained. It can be understood that the digitalization level of each level can be obtained by taking the average value of the digitalization levels of the next level, or by weighted calculation according to different preset weights.

[0056] After obtaining the overall process digitalization level of the target enterprise, the digital level of the target enterprise can be automatically divided according to the preset rules. The preset rules include setting multiple digital levels and corresponding scoring intervals, and determining the digital level of the target enterprise according to the overall process digitalization level and the scoring interval. In this application, the digital level of the enterprise is divided into 5 levels, namely, module level, unit level, process level, network level, and ecosystem level. For example, a feasible division of the scoring interval is as follows:

[0057] 0 - 1 point: Realize the informatization of some physical resources and start using simple software to manage resources, then it is rated as "basic level";

[0058] 1 - 2 points: Realize the basic data sharing between some informatization resource objects and use basic digital tools, then it is rated as "unit level";

[0059] 2 - 3 points: If the initial digitization of the application process is achieved and the process is optimized with software assistance, it is rated as "Integration Level".

[0060] 3 - 4 points: If the complete digitization of the application process is achieved and digital technologies can be integrated with actual business, it is rated as "Network Level".

[0061] 4 - 5 points: If the initial intelligence of the application process is achieved and the workflow is improved and optimized by introducing automation tools, artificial intelligence, or machine learning technologies, it is rated as "Ecosystem Level".

[0062] On the other hand, the overall score can also be obtained by comparing the existing object data in the target enterprise with the industry benchmark values in the matched industry knowledge graph. In this application, the industry benchmark values are specifically calculated based on the data of the top 10% of enterprises in the industry. The deviation between the existing object data in the target enterprise and the industry benchmark values can be evaluated, and the existing object data in the target enterprise can be quantitatively scored in combination with an intelligent algorithm. A feasible approach is to use a dynamic baseline algorithm to calculate the corresponding process - level score according to the existing object data in the device layer, edge layer, and enterprise layer in each process, and then obtain the scenario - level score, type - level score, and link - level score in sequence, so as to obtain the final overall score. The industry benchmark values are dynamically adjusted, and the benchmark values can be updated at preset times (for example, adjusted quarterly).

[0063] The above has obtained the digital levels based on the process dimension and the overall score based on the object dimension respectively. Further, the data of the above two dimensions can be combined. Specifically, a random forest algorithm (including process - level coding, object - dimension features, and outputting the influence weights of each factor) is used to perform weighted calculation on the evaluation results of the process and object dimensions, and the comprehensive maturity level is output (for example, Integration Level - 65 points).

[0064] Step 5: Generate an intelligent diagnosis report.

[0065] An intelligent diagnosis report is automatically generated based on the above results, which can at least include: digital stage positioning and visualization radar chart.

[0066] Embodiment 2

[0067] The present invention provides an enterprise digital diagnosis system based on process - object dual - drive and knowledge graph. This system is established based on the method in Embodiment 1, and the system includes a data acquisition module, an industry knowledge graph module, and a dual - drive evaluation module.

[0068] A data collection module for collecting digital transformation information of target enterprises. The digital transformation information may include: basic information (such as the industry type, scale, main business scenarios, etc. of the target enterprise) and business process data, etc. It can be understood that the digital transformation information can be imported manually or the data collection module can access the data interface provided by the target enterprise to perform automatic collection. The collection process can refer to that described in Embodiment 1. The data collection module can also obtain the existing object data of each process in the target enterprise according to the collected digital transformation information.

[0069] The industry knowledge graph module includes an industry knowledge graph constructed based on industry knowledge. The construction of the industry knowledge graph generally includes processes such as data collection, data preprocessing, knowledge extraction, ontology construction, knowledge fusion, and knowledge storage. In this application, the industry knowledge graph is managed through the graph database Neo4j. It can be understood that the industry knowledge graph can include knowledge of different sub - industries, that is, there are corresponding sub - industry knowledge graphs for different sub - industries, and the sum of all sub - industry knowledge graphs is the industry knowledge graph.

[0070] The industry knowledge graph module can also perform dynamic association optimization. This includes dynamically associating the digital levels of enterprises in this industry in links such as design, production, supply chain, sales, and service, deploying a graph neural network (GNN) for relationship reasoning; using a reinforcement learning framework (PPO algorithm) to optimize node priorities; analyzing new enterprise data through a large model (Transformer). On the one hand, constructing a LoRA fine - tuning layer and injecting enterprise real - time data (such as device logs, work order records, etc.) into the LLaMA - 2 model; on the other hand, identifying key features through attention mechanism visualization (Transformer - Explain tool) and automatically optimizing the weights of graph nodes, including indicators such as node connection entropy and relationship density.

[0071] The industry knowledge graph module will match the sub - industry knowledge graph to which the target enterprise belongs according to the digital transformation information of the target enterprise. The matching of the sub - industry knowledge graph is implemented using an NLP matching algorithm, which can automatically recommend the most relevant sub - industry knowledge graph, and the confidence level of the recommendation result can reach more than 85%.

[0072] The industry knowledge graph module can also score by comparing the existing object data of the target enterprise with the industry benchmark values in the corresponding knowledge graph. The specific scoring method can refer to that described in Embodiment 1.

[0073] The dual-drive evaluation module is used to evaluate the target enterprise in terms of both process and object dimensions. It can be understood that the dual-drive evaluation module is constructed based on the process-object evaluation system in Embodiment 1. In this module, multiple processes obtained by dividing the production and operation activities of enterprises in different industries and the corresponding object sets for each process are preset, and each object set includes several objects. As described in Embodiment 1, the production and operation activities of an enterprise can be divided into five links: design, production, supply chain, sales, and service. Each link can be divided into several types, each type can be divided into several scenarios, and each scenario can be divided into several processes.

[0074] The dual-drive evaluation module receives the existing object data in the target enterprise from the data collection module, and compares the existing object data of each process with the corresponding object set of this process to evaluate the digitalization degree of each process. The digitalization degree of each scenario can also be obtained based on the digitalization degree of each process, and then the digitalization degree of each type and each link can be obtained in turn, and finally the overall process digitalization degree of the target enterprise can be obtained. The dual-drive evaluation module will automatically divide the digitalization level of the target enterprise according to the overall process digitalization degree of the target enterprise and the preset rules. The digitalization levels include module level, unit level, process level, network level, and ecological level. The specific determination logic can refer to that described in Embodiment 1.

[0075] It also includes a diagnostic report generation module for generating a diagnostic report. The diagnostic report generation module can perform weighted calculation on the process and object dimension scores based on the processing results of the industry knowledge graph module and the dual-drive evaluation module, and output the comprehensive maturity level. The diagnostic report can at least include: digitalization stage positioning and visualization radar chart.

[0076] Some common English nouns or letters used in the present invention for the convenience of clear description are only for exemplary reference rather than restrictive interpretation or specific usage, and the protection scope of the present invention should not be limited by their possible Chinese translations or specific letters.

[0077] It should also be noted that in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

Claims

1. An enterprise digital diagnosis method based on process-object dual-drive and knowledge graph, characterized in that Including: Construct an industry knowledge graph based on industry knowledge, where the industry knowledge graph includes several sub-industry knowledge graphs; For each sub-industry, divide the business processes of an enterprise into multiple processes, and correspondingly set an object set for each process. The object set includes at least one object. Using the process as the basic evaluation item and the object as the corresponding evaluation index, form an evaluation system based on the dual drive of process-object; Collect the digital transformation information of the target enterprise, and match the corresponding sub-industry knowledge graph and evaluation system for the target enterprise; Based on the preset processes and object sets in the matched evaluation system, extract the object data that already exists in the target enterprise for each process from the digital transformation information; Based on the relationship between the object data that already exists in the target enterprise and the object set, divide the digital level of the target enterprise according to the preset rules; Based on the relationship between the object data that already exists in the target enterprise and the industry benchmark value in the sub-industry knowledge graph, obtain the overall score; Based on the digital level and the overall score, obtain the comprehensive maturity level of the target enterprise.

2. The enterprise digital diagnosis method based on process-object dual-drive and knowledge graph according to claim 1, wherein The construction process of the industry knowledge graph includes: Obtain data from the industry knowledge base; Clean the data obtained from the industry knowledge base; Extract <entity, relationship, attribute> triples from the cleaned data; Construct a domain ontology covering 5 major links of design, production, supply chain, sales and service.

3. The enterprise digital diagnosis method based on process-object dual drive and knowledge graph according to claim 2, wherein The industry knowledge graph dynamically associates the digital levels of enterprises in the sub-industry in 5 links of design, production, supply chain, sales and service, and deploys a graph neural network for relationship reasoning.

4. A method for enterprise digital diagnosis based on process-object dual drive and knowledge graph according to claim 2, characterized in that, The industry knowledge graph will be iteratively updated based on the newly added enterprise data, including: Construct a LoRA fine-tuning layer and inject the enterprise data into the LLaMA-2 model; Visually identify key features through the attention mechanism and automatically optimize the node weights of the industry knowledge graph.

5. The enterprise digital diagnosis method based on process-object dual drive and knowledge graph according to claim 1, characterized in that The digital transformation information includes basic information and business process data; The basic information includes at least the industry type, scale and main business scenarios of the target enterprise; The business process data includes device layer data, edge layer data and enterprise layer data.

6. The enterprise digital diagnosis method based on process-object dual drive and knowledge graph according to claim 5, characterized in that Match the corresponding sub-industry knowledge graph and evaluation system for the target enterprise based on the basic information; The matching of the sub-industry knowledge graph is implemented using an NLP matching algorithm.

7. A method for enterprise digital diagnosis based on process-object dual drive and knowledge graph according to claim 1, characterized in that Calculate the overall digital degree of the processes of the target enterprise based on the ratio of the number of object data that already exists in the target enterprise to the number of objects included in the object set, and divide the digital level of the target enterprise based on the overall digital degree of the processes and the preset rules; The preset rules include setting five levels of digital levels and corresponding score intervals, and determining the digital level of the target enterprise according to the overall digital degree of the processes and the score intervals.

8. A method for enterprise digital diagnosis based on process-object dual drive and knowledge graph according to claim 1, characterized in that The industry benchmark value is calculated based on the data of the top 10% of enterprises in the industry; Conduct a quantitative score based on the deviation between the object data that already exists in the target enterprise and the industry benchmark value to obtain the overall score.

9. The enterprise digital diagnosis method based on process-object dual drive and knowledge graph according to claim 1, characterized in that Based on the digital level and the overall score, a weighted calculation is performed using the random forest algorithm to obtain the comprehensive maturity level of the target enterprise.

10. An enterprise digital diagnosis system based on process-object dual-drive and knowledge graph, characterized in that, The system includes a data collection module, an industry knowledge graph module, and a dual-drive evaluation module; The industry knowledge graph module includes an industry knowledge graph constructed based on industry knowledge, and the industry knowledge graph includes a number of sub-industry knowledge graphs; The dual-drive evaluation module includes an evaluation system based on process-object dual-drive for each sub-industry. The evaluation system based on process-object dual-drive includes multiple processes obtained by subdividing the business processes of an enterprise and an object set corresponding to each of the processes. The object set includes at least one object; The data collection module is used to collect the digital transformation information of the target enterprise and extract the object data existing in the target enterprise for each process; Based on the digital transformation information, the dual-drive evaluation module matches a corresponding evaluation system for the target enterprise, and determines the digital level of the target enterprise based on the relationship between the existing object data in the target enterprise and the object set; Based on the digital transformation information, the industry knowledge graph module matches a corresponding sub-industry knowledge graph for the target enterprise, and determines the overall score based on the relationship between the existing object data in the target enterprise and the industry benchmark value in the sub-industry knowledge graph; It further includes a diagnostic report generation module for generating a diagnostic report according to the processing results of the industry knowledge graph module and the dual-drive evaluation module.

Images