Enterprise digital diagnosis system and method based on large model technology

Abstract

The invention discloses an enterprise digital diagnosis system and method based on a large model technology, and belongs to the technical field of enterprise digital diagnosis, and the system comprises an enterprise filing module, an enterprise data analysis module, an intelligent scoring module and an automatic report generation module. According to the method, strong data analysis capability and text generation capability are fully utilized, the AI technology is used for concluding and analyzing field investigation data, the digital level of the enterprise is evaluated according to related evaluation standards, a targeted solution is provided for the problems of pain points and difficulty in the digital transformation process of the enterprise, and the method is suitable for popularization and application. And a complete digital diagnosis report is issued by one key, so that the quality and the efficiency of digital diagnosis service are improved, repeated work in a manual diagnosis process is reduced, and the labor cost is effectively saved.

Description

Technical Field

[0001] The present invention relates to the technical field of enterprise digital diagnosis, and in particular to an enterprise digital diagnosis system and method based on large model technology. Background Art

[0002] The existing enterprise digital diagnosis system adopts the form of online questionnaire diagnosis filled out by enterprise information personnel or diagnostic personnel. The platform presets the scoring criteria, matches the user's selected options with the scores, and uses the corresponding weights for weighted calculation to obtain the scores and situation analysis reports related to the enterprise's digital level. This method is limited by the evaluator's understanding of the enterprise's digital level and the rationality of the questionnaire setting, and may not be able to accurately reflect the actual situation of the enterprise. And it cannot propose customized solutions for the specific pain points and difficulties of the enterprise.

[0003] Manual diagnosis has higher reliability than online platform diagnosis. However, traditional manual diagnosis has limitations. The diagnostic process involves a lot of repetitive work in text data processing and text writing, which not only consumes a lot of manpower and material resources, but also makes it difficult to ensure the accuracy and timeliness of the diagnosis results. This not only affects the quality of diagnosis, but may also bring unnecessary risks to the enterprise. To this end, a digital enterprise diagnosis system and method based on large model technology is proposed. Summary of the invention

[0004] The technical problem to be solved by the present invention is: how to improve the quality and efficiency of enterprise digital diagnosis services, reduce repetitive work in the manual diagnosis process, effectively save labor costs, and provide an enterprise digital diagnosis system based on large model technology.

[0005] The present invention solves the above technical problems through the following technical solutions, which include an enterprise file building module, an enterprise data analysis module, an intelligent scoring module and an automatic report generation module;

[0006] The enterprise file building module is used to analyze the enterprise basic information document, obtain key information in the enterprise basic information document, and build an independent enterprise file;

[0007] The enterprise data analysis module is used to analyze the on-site survey data of the enterprise to obtain analysis results;

[0008] The scoring module is used to score the digitalization level of the enterprise according to the relevant evaluation model to obtain the scoring result;

[0009] The report generation module is used to generate an enterprise digital diagnosis report based on the analysis results of the enterprise data analysis module by relying on large model technology.

[0010] Furthermore, in the enterprise file creation module, the key information in the enterprise basic information document includes the enterprise name, industry, enterprise scale, industry, and annual output value.

[0011] Furthermore, in the enterprise data analysis module, the formats of the field survey data include text, tables, slides, images, and audio.

[0012] Furthermore, in the enterprise data analysis module, the field survey data is analyzed based on large model technology and multiple machine learning algorithms, as follows:

[0013] For text data, a pre-trained semantic large model is used for analysis;

[0014] For tabular data, the extracted tabular data is converted into a format that can be processed by the large model, the tabular data is structured and stored, the table is semantically processed, the tabular information is converted into semantic text, the semantic text is backfilled to the corresponding position of the document according to special placeholders, the tabular data is converted into text data, and then the pre-trained semantic large model is used for analysis;

[0015] For slide data, we use the target detection algorithm to detect all the text content and location information of the page, use the OCR algorithm to identify the text information in the text content, and combine the detected location information to store the text content in a structured manner; perform semantic processing on the text content to convert it into semantic text, and input the semantic text into the pre-trained semantic large model for analysis;

[0016] For image data, we use target detection algorithms to identify objects in input images, and use causal reasoning algorithms to understand image scenes. We output the semantic information of the entire image, save the semantic information as text data, and input it into the pre-trained semantic model for analysis.

[0017] For audio data, audio content is obtained from the microphone or read from the storage device and normalized and denoised, the voice content is recognized and converted into text using a speech-to-text algorithm, and the text is input into a pre-trained semantic large model for analysis.

[0018] Furthermore, in the scoring module, the specific steps of training and scoring are:

[0019] S31: Sort out the scoring items and corresponding options according to the intelligent manufacturing capability maturity evaluation criteria, and assign scoring weights to the scoring items according to their importance;

[0020] S32: Build a model training set and score the enterprise's digitalization level based on the scoring items and the analysis results of the enterprise data analysis module;

[0021] S33: training the big model scoring function, inputting the analysis results of the enterprise data analysis module into the big model for training, and saving the big model with the most accurate evaluation results;

[0022] S34: Use the trained big model to score the digitalization level of the enterprise and present the scoring results in a table.

[0023] Furthermore, in step S31, the evaluation criteria include four elements, namely personnel, technology, resources, and manufacturing. The four elements are further subdivided into multiple capability domains. The personnel element is subdivided into two capability domains, including organizational strategy and personnel skills; the technology element is subdivided into three capability domains, including data, integration, and information security; the resource element is subdivided into two capability domains, including equipment and network; the manufacturing element is subdivided into five capability domains, including design, production, logistics, sales, and service. Each capability domain is further subdivided into one or more capability subdomains, and each capability subdomain is provided with multiple scoring items, and each scoring item corresponds to multiple options.

[0024] Furthermore, in step S32, the analysis results of the enterprise data analysis module include a description of the enterprise's current status in terms of organizational strategy, personnel skills, data, integration, information security, equipment, network, design, production, logistics, sales, and services; in step S34, the trained large model can match the scoring items and options corresponding to the scoring criteria according to the keywords in the analysis results of the above-mentioned enterprise data analysis module, thereby achieving scoring.

[0025] Furthermore, in the report generation module, according to the report framework set by the user, based on the analysis results of the enterprise data analysis module, and combined with the corresponding content in the knowledge base, a diagnostic report is generated, and the scoring results in tabular form are simultaneously reflected in the diagnostic report.

[0026] Furthermore, the knowledge base includes enterprise digital transformation solutions, industry benchmark cases, transformation general contracting service providers and excellent solution service providers.

[0027] The present invention also provides a method for digital diagnosis of an enterprise based on a large model technology, comprising the following steps:

[0028] S1: Enterprise file creation

[0029] Users upload enterprise basic information documents, analyze the enterprise basic information documents, obtain key information in the enterprise basic information documents, and build independent enterprise files;

[0030] S2: Data Analysis

[0031] Analyze the on-site survey data of the enterprise and obtain the analysis results;

[0032] S3: Diagnostic scoring

[0033] Based on the analysis results and according to the relevant evaluation model, the digitalization level of the enterprise is scored to obtain the scoring results;

[0034] S4: Diagnostic report generation

[0035] Used to generate an enterprise digital diagnosis report based on the analysis results in step S2 by relying on large model technology.

[0036] Compared with the existing technology, the present invention has the following advantages: the enterprise digital diagnosis system and method based on large model technology makes full use of powerful data analysis capabilities and text generation capabilities, uses AI technology to summarize and analyze on-site survey data, and evaluates the digitalization level of the enterprise according to relevant evaluation standards. It proposes targeted solutions to the pain points and difficulties in the process of enterprise digital transformation, and issues a complete digital diagnosis report with one click, aiming to improve the quality and efficiency of digital diagnosis services, reduce repetitive work in the manual diagnosis process, and effectively save labor costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a flow chart of the enterprise digital diagnosis method based on the large model technology of the present invention;

[0038] Figure 2 is a schematic diagram of the evaluation standard system structure in an embodiment of the present invention;

[0039] FIG3( a ) is a schematic diagram showing the effect of background information on output not being eliminated in an embodiment of the present invention;

[0040] FIG. 3( b ) is a schematic diagram of eliminating the influence of background information on output according to an embodiment of the present invention. DETAILED DESCRIPTION

[0041] The following is a detailed description of an embodiment of the present invention. This embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation method and a specific operation process are given, but the protection scope of the present invention is not limited to the following embodiment.

[0042] Embodiment 1

[0043] This embodiment provides a technical solution: a digital enterprise diagnosis system based on large model technology, including an enterprise file building module, an enterprise data analysis module, an intelligent scoring module and an automatic report generation module;

[0044] The enterprise file building module is used to analyze the basic information documents of the enterprise. Based on the large model text understanding ability, it automatically obtains the key information contained in the basic information documents of the enterprise, such as the enterprise name, industry, enterprise scale, industry, annual output value, etc., and builds an independent enterprise file. The key information collected here is the basic information of the enterprise, which is used to build the enterprise file. After the report is generated, the above information will be displayed first when the user queries again.

[0045] Enterprise data analysis module, which is used to deeply analyze various forms of enterprise digital application related information. Based on big model technology and multiple machine learning algorithms, it can analyze and process field survey data in various forms such as text, tables, slides, images, audio, etc., and format and store them;

[0046] More specifically, the process of analyzing the above-mentioned various types of data in this embodiment is as follows:

[0047] For text data, a pre-trained semantic large model is used for analysis and processing;

[0048] For tabular data, S1: the extracted tabular data needs to be converted into a format that can be processed by the large model, and the tabular data needs to be stored in a structured manner, such as storing each column of data in an array, converting multiple one-dimensional arrays into two-dimensional arrays, or converting the data into a DataFrame data frame; S2: semantic processing of the table to convert the table information into semantic text; S3: backfilling the semantic text to the corresponding position of the document according to the special placeholder. In this way, the tabular information is converted into text information, which can be further analyzed and processed by the pre-trained semantic large model.

[0049] For image data, S1: Use target detection algorithm (such as YOLO, Fast-RCNN) to identify objects in the input image; S2: Use causal reasoning algorithm to understand the image scene and output the semantic information of the whole image; S3: Save the above semantic information as text input into the pre-trained semantic model for analysis and processing.

[0050] In this embodiment, the purpose of the causal inference algorithm here is: when the model extracts high-dimensional features from an image, it pays more attention to the causal information corresponding to the result, and tries to ignore the false correlation between the background and the target. An abstract example is that the input is causal information X and background information C, and V represents the common result of the causal information and the background information, as shown in Figure 3(a). The purpose of causal inference is to eliminate the influence of the false correlation of a certain type of background information on X on the output Y, as shown in Figure 3(b).

[0051] For the slide data, S1: use the target detection algorithm to detect all the text content and location information of the page; S2: use the OCR algorithm to recognize the text information, and combine the location information in step S1 to store the text content in a structured manner; S3: perform semantic processing on the data obtained in step S2 to convert the text content into semantic text. S4 inputs the text into the pre-trained semantic large model for analysis.

[0052] For audio data, S1: obtains audio content from a microphone or reads audio from a storage device and performs normalization and noise reduction processing; S2: recognizes voice content and uses speech-to-text algorithms such as Whisper to convert speech into text; S3 inputs the text into a large language model for analysis.

[0053] The intelligent scoring module scores the digitalization level of enterprises according to relevant evaluation models (intelligent manufacturing capability maturity model, small and medium-sized enterprise digital transformation evaluation standards, etc.) and obtains the scoring results;

[0054] The processing steps of the intelligent scoring module are:

[0055] Step 1: Sort out scoring items and corresponding options according to evaluation criteria such as intelligent manufacturing capability maturity standards, and assign scoring weights to them according to the importance of the items;

[0056] like Figure 2 As shown in the table, the evaluation criteria include personnel (P), technology (T), resources (R) and manufacturing (M), which are further divided into 12 capability domains and 20 capability subdomains. There are more than 300 scoring items and more than 800 options involved.

[0057] The following are examples of scoring items and corresponding options:

[0058] Scoring items: Data management, belonging to the capability subdomain - data;

[0059] The corresponding options are as follows:

[0060] Option A: Have a data management plan;

[0061] Option B: It can transmit and store the massive data generated in the production and operation management process, but cannot perform deeper data processing;

[0062] Option C: It can clean, integrate, and visualize massive amounts of data on the basis of transmission and storage;

[0063] Option D: Initially deploy and build the cloud within the factory, establish a database based on cloud computing architecture, and perform data management work in the cloud;

[0064] Option E: Data storage, preprocessing, and other tasks can be performed at the network edge, close to the device or data source, and coordinated with data management in the cloud.

[0065] Scoring items: The company's ability to understand and grasp customer needs during the product life cycle, which belongs to the capability subdomain - product service;

[0066] The corresponding options are as follows:

[0067] Option A: Not involved, start planning for customer demand insights;

[0068] Option B: When the product is in the early design and planning stage, you can accurately grasp the customer's needs for product and service features and selling points, and transform the needs into the product design process; for products that have already been launched, adjust and optimize the products and services by analyzing the gap between the current product features and meeting customer needs;

[0069] Option C: Build a product demand analysis model based on the usage or experience data from customer feedback. The analysis results based on the model can identify problems in product use.

[0070] Option D: Based on product attributes, sales, and usage data, establish a product demand forecasting model, perform clustering and data mining, and correctly forecast customers’ potential product demand;

[0071] Option E: By combining the needs for after-sales service response, after-sales service convenience, and new product features, product innovation solutions can be optimized in real time.

[0072] Step 2: Build a model training set and score the enterprise's digitalization level based on the scoring items and the analysis results of the enterprise data analysis module;

[0073] In this embodiment, the analysis results of the enterprise data analysis module mainly include the current status of the enterprise in terms of organizational strategy, personnel skills, data, integration, information security, equipment, network, design, production, logistics, sales, and services. For example, the enterprise has fiber-optic broadband, which can meet the company's daily office, data transmission, and possible online business communication and other network needs. The company currently uses the agency accounting method for financial affairs, and has not yet put other information systems online.

[0074] In this embodiment, the big model will match the keywords of the above analysis results with the corresponding items and options of the scoring criteria. For example, in the analysis results, the company's current financial accounting method is adopted, and no other information systems are online. The relevant items of the information system are matched. Since only the financial system is online and there are no other systems, the corresponding items of the financial system here are scored, and the remaining information systems are all 0 points.

[0075] In this embodiment, the specific content, scoring items, options, and scores corresponding to the content in the early training are determined and labeled by professionals by analyzing various types of survey data.

[0076] Step 3: Train the big model scoring function, input the analysis results of the enterprise data analysis module into the big model for training, and save the model with the most accurate evaluation results;

[0077] Step 4: Use the trained model to score the digitalization level of the enterprise and present the scoring results in a table.

[0078] The automatic report generation module is used to automatically generate enterprise digital diagnosis reports based on the analysis results of the enterprise data analysis module by relying on large model technology.

[0079] The scoring results in tabular form are displayed in the system interface in a tabular form to show the scores of the above 20 capability sub-domains of the enterprise.

[0080] The method of automatically generating diagnostic reports is based on the diagnostic report knowledge base built by the diagnostic service provider, which contains relevant knowledge such as enterprise digital transformation solutions, industry benchmark cases, transformation general contractors, and excellent solution providers. According to the report framework set by the user, based on the analysis results of the enterprise data analysis module and combined with the corresponding content in the knowledge base, the diagnostic report is intelligently generated, and the scoring results in the form of a table are simultaneously reflected in the diagnostic report.

[0081] like Figure 1 As shown, this embodiment also provides an enterprise digital diagnosis method based on large model technology, using the above-mentioned enterprise digital diagnosis system to perform enterprise digital diagnosis work, including the following steps:

[0082] S1: Enterprise file creation

[0083] Users upload a document containing basic information about the company, and the system intelligently extracts basic information such as the company name, industry, company size, industry, and annual output value contained in the document;

[0084] S2: Field survey data analysis

[0085] Determine whether to collect on-site research recordings. After collecting the recordings, they can be intelligently transcribed into text analysis. If there is no need to collect, you can directly upload on-site research materials, such as images, videos, texts, PPTs and other materials for analysis. The system is based on the large model data analysis capabilities to capture the pain points and difficulties in the digital transformation process of enterprises;

[0086] S3: Diagnostic scoring: Based on the intelligent manufacturing capability maturity assessment standards and data analysis results, the digitalization level of the enterprise is intelligently scored and rated. On-site diagnostic experts can manually calibrate the scores.

[0087] S4: Diagnostic report generation, relying on the large model text generation capability, generates a digital enterprise diagnostic report with both pictures and texts, and can use dialogue interaction to modify the diagnostic report. After confirming the diagnostic report, the report can be exported to the local computer.

[0088] To sum up, the enterprise digital diagnosis system and method based on big model technology in the above-mentioned embodiment can intelligently analyze and summarize various forms of research data obtained at the diagnosis and investigation site, which can effectively reduce the time for diagnostic technicians to organize and analyze data and efficiently obtain effective information; it can combine relevant standards, understand and analyze the pain points and difficulties in the process of enterprise digital transformation, and provide customized solutions for these problems, and generate complete digital diagnosis reports. At the same time, the system also has powerful industry analysis capabilities, which can statistically analyze the common needs of the industry, provide solutions, and explore subsequent business opportunities.

[0089] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. An enterprise digital diagnosis system based on large model technology, characterized in that: include: Enterprise file building module, enterprise data analysis module, intelligent scoring module and automatic report generation module; The enterprise file building module is used to analyze the enterprise basic information document, obtain key information in the enterprise basic information document, and build an independent enterprise file; The enterprise data analysis module is used to analyze the on-site survey data of the enterprise to obtain analysis results; The scoring module is used to score the digitalization level of the enterprise according to the relevant evaluation model to obtain the scoring result; The report generation module is used to generate an enterprise digital diagnosis report based on the analysis results of the enterprise data analysis module by relying on large model technology.

2. According to claim 1, a digital enterprise diagnosis system based on large model technology is characterized in that: In the enterprise file creation module, the key information in the enterprise basic information document includes the enterprise name, industry, enterprise scale, industry, and annual output value.

3. According to claim 2, a digital enterprise diagnosis system based on large model technology is characterized in that: In the enterprise data analysis module, the formats of field survey data include text, tables, slides, images, and audio.

4. According to claim 3, a digital enterprise diagnosis system based on large model technology is characterized in that: In the enterprise data analysis module, the field survey data is analyzed based on large model technology and multiple machine learning algorithms, as follows: For text data, a pre-trained semantic large model is used for analysis; For tabular data, the extracted tabular data is converted into a format that can be processed by the large model, the tabular data is structured and stored, the table is semantically processed, the tabular information is converted into semantic text, the semantic text is backfilled to the corresponding position of the document according to special placeholders, the tabular data is converted into text data, and then the pre-trained semantic large model is used for analysis; For slide data, the target detection algorithm is used to detect all the text content and location information of the page, and the OCR algorithm is used to identify the text information in the text content. Combined with the detected location information, the text content is stored in a structured manner; Perform semantic processing on the text content to convert it into semantic text, and input the semantic text into the pre-trained semantic model for analysis; For image data, we use target detection algorithms to identify objects in input images, and use causal reasoning algorithms to understand image scenes. We output the semantic information of the entire image, save the semantic information as text data, and input it into the pre-trained semantic model for analysis. For audio data, audio content is obtained from the microphone or read from the storage device and normalized and denoised, the voice content is recognized and converted into text using a speech-to-text algorithm, and the text is input into a pre-trained semantic large model for analysis.

5. According to claim 1, the enterprise digital diagnosis system based on large model technology is characterized in that: In the scoring module, the specific steps of training and scoring are: S31: Sort out the scoring items and corresponding options according to the intelligent manufacturing capability maturity evaluation criteria, and assign scoring weights to the scoring items according to their importance; S32: Build a model training set and score the enterprise's digitalization level based on the scoring items and the analysis results of the enterprise data analysis module; S33: training the big model scoring function, inputting the analysis results of the enterprise data analysis module into the big model for training, and saving the big model with the most accurate evaluation results; S34: Use the trained big model to score the digitalization level of the enterprise and present the scoring results in a table.

6. The enterprise digital diagnosis system based on large model technology according to claim 5 is characterized in that: In step S31, the evaluation criteria include four elements, namely personnel, technology, resources, and manufacturing. The four elements are further subdivided into multiple capability domains. The personnel element is subdivided into two capability domains, including organizational strategy and personnel skills. The technical element is subdivided into three capability domains, including data, integration, and information security. The resource element is subdivided into two capability domains, including equipment and network. The manufacturing element is subdivided into five capability domains, including design, production, logistics, sales, and service. Each capability domain is further subdivided into one or more capability subdomains. Each capability subdomain is provided with multiple scoring items, and each scoring item corresponds to multiple options.

7. The enterprise digital diagnosis system based on large model technology according to claim 5 is characterized in that: In step S32, the analysis results of the enterprise data analysis module include a description of the enterprise's current status in terms of organizational strategy, personnel skills, data, integration, information security, equipment, network, design, production, logistics, sales, and services; in step S34, the trained large model can match the scoring items and options corresponding to the scoring criteria according to the keywords in the analysis results of the above-mentioned enterprise data analysis module, thereby achieving scoring.

8. The enterprise digital diagnosis system based on large model technology according to claim 1 is characterized in that: In the report generation module, according to the report framework set by the user, based on the analysis results of the enterprise data analysis module and combined with the corresponding content in the knowledge base, a diagnostic report is generated, and the scoring results in tabular form are simultaneously reflected in the diagnostic report.

9. The enterprise digital diagnosis system based on large model technology according to claim 8 is characterized in that: The knowledge base includes enterprise digital transformation solutions, industry benchmark cases, transformation general contracting service providers and excellent solution service providers.

10. A digital enterprise diagnosis method based on large model technology, characterized in that: The enterprise digital diagnosis system according to any one of claims 1 to 9 is used to perform enterprise digital diagnosis work, and the steps include: S1: Enterprise file creation Users upload enterprise basic information documents, analyze the enterprise basic information documents, obtain key information in the enterprise basic information documents, and build independent enterprise files; S2: Data Analysis Analyze the on-site survey data of the enterprise and obtain the analysis results; S3: Diagnostic scoring Based on the analysis results and according to the relevant evaluation model, the digitalization level of the enterprise is scored to obtain the scoring results; S4: Diagnostic report generation Used to generate an enterprise digital diagnosis report based on the analysis results in step S2 by relying on large model technology.

Images