An AI-based enterprise diagnostic report generation method and system

Abstract

The present invention proposes an AI-based enterprise diagnosis report generation method and system, which relate to the field of enterprise management technology. The method includes dividing different industries, dividing the key steps of the business processes of each industry and setting process indicators, obtaining ranking weights of different types of key steps, ranking the process indicators by importance through the ranking weights, and adding classification dimensions to build a structured indicator library; through the indicator library, setting evaluation questionnaires for different industries and the number of questions for different types of key steps, and implanting diagnostic expert evaluation logic; setting a variety of diagnostic models, and selecting a suitable diagnostic model according to the matching degree between the content of the evaluation questionnaire and the model and the recognition of the model by different cities; collecting enterprise data and diagnostic models through the evaluation questionnaire to obtain diagnostic results; generating diagnostic reports through AI, which not only improves the efficiency and accuracy of data processing, but also provides customized diagnostic reports.

Description

Technical Field

[0001] The present application relates to the field of enterprise management technology, and in particular to an AI-based enterprise diagnostic report generation method and system. Background Art

[0002] In the current business environment, companies need to constantly review and optimize their operational strategies in order to remain competitive. However, corporate diagnosis is often a complex and time-consuming process involving in-depth analysis and evaluation of multiple business processes. Traditional corporate diagnosis methods rely on manual review and data collection, which is not only inefficient but also susceptible to subjective factors, which affects the accuracy and reliability of diagnostic results.

[0003] With the rapid development of artificial intelligence (AI) technology, its application in the field of enterprise diagnosis has become more and more extensive. AI technology can automatically process and analyze large amounts of data and provide fast and accurate diagnosis results. However, existing AI-based enterprise diagnosis methods still have some shortcomings.

[0004] For example, these methods usually lack customized solutions for different industries and business models, resulting in insufficient pertinence and practicality of diagnostic results. In addition, these methods also have certain limitations in data collection and processing, such as incomplete data, data anomalies, etc., which may affect the accuracy of diagnostic results; when building an indicator library, how to ensure the comprehensiveness and accuracy of indicators; when setting up evaluation questionnaires, how to formulate scientific and reasonable questionnaire content and number of questions according to the characteristics of different industries and key steps; when selecting a diagnostic model, how to match the questionnaire content with the model characteristics, and how to deal with the situation where the model is not well matched; when generating a diagnostic report, how to combine elements such as pictures and charts to improve the readability and guidance of the report, etc., are all technical problems that need to be solved urgently. Summary of the invention

[0005] Based on the above problems, this application provides an AI-based enterprise diagnostic report generation method and system. By combining the funnel model, cluster analysis, and database statistical data analysis technology, a structured indicator library is constructed and sorted according to the importance of different industries and key steps; appropriate diagnostic models are selected to meet the needs of different cities and enterprises; not only the efficiency and accuracy of data processing are improved, but also customized diagnostic reports can be provided according to the actual situation of different enterprises.

[0006] The purpose of this application is achieved by the following technical solutions:

[0007] In a first aspect, the present application provides an AI-based enterprise diagnostic report generation method, the method comprising:

[0008] S1. Different industries are divided according to business background and business model. The key steps of business processes in each industry are divided through the funnel model and process indicators are set. The ranking weights of different types of key steps are obtained by cluster analysis and database statistics. The importance of process indicators is ranked by ranking weights, and classification dimensions are added to build a structured indicator library.

[0009] S2. Using the indicator library, set evaluation questionnaires for different industries and the number of questions for different types of key steps, and embed diagnostic expert evaluation logic;

[0010] S3. Set up multiple diagnostic models and select the appropriate diagnostic model based on the matching degree between the evaluation questionnaire content and the model and the recognition of the model by different cities;

[0011] S4. Collect enterprise data through the evaluation questionnaire, input the enterprise data into the selected diagnosis model, and obtain the diagnosis result;

[0012] S5. Generate diagnostic reports through AI.

[0013] In a second aspect, the present application provides an enterprise diagnostic report generation system based on AI, the system comprising:

[0014] The indicator library construction module is used to divide different industries according to business background and business model, divide the key steps of the business process of each industry through the funnel model and set process indicators, use cluster analysis and database statistics to obtain the ranking weights of different types of key steps, and then rank the process indicators in importance, add classification dimensions, and build a structured indicator library;

[0015] A questionnaire setting module is used to set evaluation questionnaires for different industries and the number of questions for different types of key steps through the indicator library, and to implant diagnostic expert evaluation logic;

[0016] The model selection module is used to set up multiple diagnostic models and select the appropriate diagnostic model based on the matching degree between the evaluation questionnaire content and the model and the recognition of the model by different cities;

[0017] A diagnosis module, used to collect enterprise data through an evaluation questionnaire, input the enterprise data into a selected diagnosis model, and obtain a diagnosis result;

[0018] Report generation module, used to generate diagnostic reports through AI.

[0019] In a third aspect, the present application also provides an electronic device comprising a memory, a processor, and a computer program stored on and executable on the memory, wherein the processor executes the program to implement any AI-based enterprise diagnostic report generation method described in the present application.

[0020] The beneficial effects of the present invention include: by constructing a structured index library, the key steps of the business process of each industry are scientifically divided, and the process indicators and their importance ranking are set, which provides clear and quantitative evaluation standards for enterprise diagnosis and improves the accuracy and efficiency of diagnosis. An evaluation questionnaire is set according to the index library, and the diagnostic expert evaluation logic is implanted, which further enhances the professionalism and pertinence of the diagnosis. The step realizes the intelligent selection and optimization of the diagnostic model by comprehensively considering the matching degree between the questionnaire evaluation output and the diagnostic model demand input, as well as the recognition of the model by various industries in various cities, which not only ensures the high degree of fit between the model and the actual situation of the enterprise, but also improves the practicality and reliability of the diagnostic results. Through strict integrity, validity judgment, data formatting and encoding, invalid or abnormal data are effectively eliminated, ensuring the quality of data input into the diagnostic model. This lays a solid foundation for improving the accuracy of the diagnostic results. Through intelligent data processing flow, the efficiency of data processing is significantly improved and the labor cost is reduced. Combined with pictures, charts and animation elements, the diagnostic report or interpretation content is visually presented. This not only makes the diagnostic results more intuitive and easy to understand, but also enhances the attractiveness and persuasiveness of the report. The method also supports report interpretation in voice and / or text form, further meeting the needs and preferences of different users. The function of obtaining and analyzing user voice question information can generate or interpret diagnostic reports based on user questions, providing users with a more convenient and personalized service experience and enhancing the interactivity between users and the system.

[0021] In summary, the AI-based enterprise diagnosis report generation method has significantly improved the efficiency, accuracy and practicality of enterprise diagnosis through a series of innovative designs and practices, providing enterprises with more comprehensive and accurate business assessments and optimization suggestions. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a schematic diagram of an AI-based enterprise diagnostic report generation method provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] Below, the present application is further described in conjunction with the accompanying drawings and specific implementation methods. It should be noted that, under the premise of no conflict, the various embodiments or technical features described below can be arbitrarily combined to form a new embodiment.

[0024] The present application provides an AI-based enterprise diagnostic report generation method, the method comprising:

[0025] S1. Different industries are divided according to business background and business model. The key steps of business processes in each industry are divided through the funnel model and process indicators are set. The ranking weights of different types of key steps are obtained by cluster analysis and database statistics. The importance of process indicators is ranked by ranking weights, and classification dimensions are added to build a structured indicator library.

[0026] S2. Using the indicator library, set evaluation questionnaires for different industries and the number of questions for different types of key steps, and embed diagnostic expert evaluation logic;

[0027] S3. Set up multiple diagnostic models and select the appropriate diagnostic model based on the matching degree between the evaluation questionnaire content and the model and the recognition of the model by different cities;

[0028] S4. Collect enterprise data through the evaluation questionnaire, input the enterprise data into the selected diagnosis model, and obtain the diagnosis result;

[0029] S5. Generate diagnostic reports through AI.

[0030] The working principle of the above technical solution is: first, the market is divided into different industries according to the business background and business model of the enterprise; each industry has its own unique business process and operation model; then, using the funnel model, the business process of each industry is subdivided into a series of key steps, and process indicators are set for each step; these indicators can accurately reflect the key behaviors and conversion rates in the business process, and are the basis for subsequent analysis.

[0031] In order to determine the importance of different types of key steps, cluster analysis and database statistics were used; cluster analysis grouped similar key steps into one category, while database statistics calculated their ranking weights based on the frequency and weight of these key steps in the database; this step ensured the comprehensiveness and accuracy of the indicator library and provided a basis for the design of subsequent evaluation questionnaires. When constructing the indicator library, classification dimensions such as time and region were also added to enable a more in-depth analysis of enterprise operation data.

[0032] Based on the constructed indicator library, evaluation questionnaires are designed for different industries. The questions in the questionnaire correspond to the key steps and process indicators in the indicator library to ensure that the company's operations are fully covered. At the same time, according to the ranking weights of the key steps, the number of questions corresponding to different types of key steps is adjusted to ensure the pertinence and effectiveness of the questionnaire.

[0033] In addition, diagnostic expert evaluation logic is embedded in the questionnaire, and the experts’ experience and knowledge are used to conduct preliminary analysis and judgment on the questionnaire results, thereby improving the accuracy and reliability of the diagnostic results.

[0034] In order to obtain more accurate diagnostic results, a variety of diagnostic models are set up; these models are based on different algorithms and principles and are suitable for different analysis scenarios; when selecting a suitable diagnostic model, it is necessary to consider the matching degree between the content of the evaluation questionnaire and the model, as well as the recognition of the model in different cities. At the same time, there may be differences in the recognition of the model in different cities. Therefore, when selecting a model, it is also necessary to consider urban factors to ensure that the selected model has good applicability and recognition in the local area. After collecting enterprise data through the evaluation questionnaire, these data are input into the selected diagnostic model. The model will process and analyze the data and generate diagnostic results according to preset algorithms and rules. These results include problems existing in the operation of the enterprise, suggestions for improvement, etc.; finally, the diagnostic report is automatically generated using AI technology; the report content is based on the diagnostic results and is visualized with elements such as pictures and charts; it can not only improve the readability of the report, but also more intuitively display the operation of the enterprise and the analysis results.

[0035] To summarize, the AI-based enterprise diagnostic report generation method provided in this application achieves a comprehensive analysis and diagnosis of the enterprise's operating conditions through the steps of building a structured indicator library, designing evaluation questionnaires, selecting appropriate diagnostic models, collecting data and inputting diagnostic models, and generating diagnostic reports.

[0036] The effects of the above technical solutions are as follows: by building a structured indicator library, the key steps and process indicators of enterprise operations can be fully and systematically covered to ensure the comprehensiveness of the diagnosis; cluster analysis and database statistics are used to obtain the ranking weights of key steps, and the process indicators are ranked by importance, making the diagnosis more accurate and focusing on the core issues in enterprise operations; implanting diagnostic expert evaluation logic, combined with AI technology, can automatically process a large amount of data, quickly generate diagnostic reports, and significantly improve diagnostic efficiency; according to the characteristics of different industries and key steps, set evaluation questionnaires and the number of questions, so that the diagnostic report is closer to the actual situation of the enterprise. By generating diagnostic reports through AI technology, visual elements such as pictures and charts can be combined to make the report content more intuitive and easy to understand, which is convenient for enterprises to understand and implement improvement measures. Setting up multiple diagnostic models and selecting appropriate models based on the matching degree between the content of the evaluation questionnaire and the model and the recognition of the model by different cities can ensure the accuracy and reliability of the diagnostic results. Customized diagnosis is allowed according to the characteristics of different industries and enterprises, which improves the applicability and flexibility of the diagnostic model. By using this method for enterprise diagnosis regularly or irregularly, problems in operation can be discovered and solved in a timely manner, promoting continuous improvement of enterprises.

[0037] In some embodiments, the method further comprises:

[0038] Obtain user voice question information and analyze it;

[0039] Generate a diagnostic report and / or interpret the diagnostic report based on the analysis results; interpret the report in voice and / or text form.

[0040] The working principle and effect of the above technical solution are as follows: first, through voice recognition technology, the voice question information sent by the user through a microphone or other voice input device is captured and received; this step ensures that the user can interact with the system in a natural and convenient way. The received voice information is then sent to the natural language processing (NLP) module for analysis. The NLP module can identify keywords, phrases and semantic structures in the voice and convert the voice information into a text form that can be understood by the computer. The parsing process includes multiple sub-steps such as voice to text, keyword extraction, and semantic understanding to ensure that the user's intentions and needs are accurately captured. According to the parsing results, the diagnostic report generation process is automatically triggered. If the user has submitted enterprise data through evaluation questionnaires and other methods, and the system has selected a suitable diagnostic model for diagnosis, then the diagnostic report can be directly generated and displayed; if the user's question is about how to interpret the generated diagnostic report, the key information, analysis results and improvement suggestions in the report can be intelligently extracted according to the content of the user's question, and interpreted in voice and / or text form. Through speech synthesis technology, the interpretation content is converted into voice output for users to listen to. At the same time, the system can also provide text interpretation to meet the needs of different users. Users can better understand the diagnostic report based on the interpretation provided by the system, and formulate improvement measures or make decisions accordingly. If users need further consultation or more information, manual customer service or other support channels can also be provided so that users can get more in-depth help.

[0041] To sum up, by adding the function of processing and responding to user voice question information, the AI-based enterprise diagnostic report generation method provided in this application not only improves the interactivity and ease of use of the system, but also provides users with a more convenient and comprehensive service experience.

[0042] In some embodiments, the S1 includes:

[0043] Different industries are divided according to business background and business model, including e-commerce, finance, education, and medical care;

[0044] Obtain the business processes of various industries, divide each business process of each industry into multiple key steps according to business logic and funnel model, and obtain process indicators for each step; indicators can accurately reflect the key behaviors and conversion rates in the business process; for example, in the e-commerce industry, the business funnel includes steps such as user browsing, adding to cart, and purchasing, and corresponding process indicators can be set for each step;

[0045] For the same industry, cluster analysis is performed on each key step to obtain the categories of key steps; the ranking weights of each key step are obtained based on the number of times each key step appears in the database of companies in the industry and the first weight of the key step in each company;

[0046] The ranking weight of the key steps is obtained by the following formula:

[0047] ;

[0048] in, is the ranking weight of the jth key step in the ith industry; is the first weight of the j-th key step in the ith industry among all the key steps of the k-th company; n is the number of times the j-th key step in the ith industry appears in different companies; m is the total category of the key steps in the ith industry;

[0049] According to the ranking weight of each key step, set the importance ranking of the process indicators of the corresponding step;

[0050] Adding classification dimensions to process indicators of various key steps in various industries, including time dimension and region dimension;

[0051] Build a structured indicator library through the database; and assign a unique code and name to each one.

[0052] The working principle of the above technical solution is: according to the business background and business model of the enterprise, the market is divided into different industries, such as e-commerce, finance, education, medical care, etc., to ensure that subsequent analysis can be carried out on groups of enterprises with similar business characteristics.

[0053] For each industry, obtain its typical business process; subdivide the business process into multiple key steps through business logic and funnel model; the funnel model helps to identify the key links from potential customers to actual transactions. Set process indicators for each key step, which can accurately reflect the key behaviors and conversion rates in the business process; for example, in the e-commerce industry, key steps may include user browsing, adding to cart, and purchasing, and each step can set process indicators such as browsing volume, adding to cart rate, and purchase conversion rate. Perform cluster analysis on key steps in the same industry and classify each key step into different categories based on business logic and similarity.

[0054] Through database statistics, calculate the number of times each key step appears in the companies in the industry (n) and the first weight of each key step in all key steps of each company; the first weight represents the importance of the key step in the overall business of a company; use the formula to calculate the ranking weight of each key step, where m is the total number of key steps in the industry; this step ensures that the ranking of the importance of key steps is based on widely recognized data in the industry. According to the ranking weight of each key step, the process indicators corresponding to each key step are ranked in importance. The higher the ranking of the key step, the more important its process indicators will be in the subsequent enterprise diagnosis.

[0055] Add classification dimensions to the process indicators of various key steps in various industries, such as time dimensions (such as month, quarter, year, etc.) and regional dimensions (such as province, city, etc.); these dimensions help to deeply analyze the differences in corporate operations at different times and regions. Through database technology, the key steps, process indicators, sorting weights, and classification dimensions of various industries obtained in the above steps are integrated to build a structured indicator library. Assign a unique code and name to each indicator in the indicator library to ensure accurate reference and identification in subsequent corporate diagnosis.

[0056] In summary, this method constructs a comprehensive and systematic structured indicator library through detailed industry division, business process analysis, key step identification and weight calculation, process indicator importance ranking, and addition of classification dimensions, providing a solid data foundation and support for the subsequent generation of enterprise diagnostic reports.

[0057] In some embodiments, the S2 includes:

[0058] For each type of industry, the key steps and process indicators are converted into questionnaire questions based on the indicator library; according to the ranking weight of the key step type, the preliminary number of questionnaire questions corresponding to each type of key step in each industry is set;

[0059] The preliminary number of key step questions is:

[0060] ;

[0061] in, is the number of questions corresponding to the jth key step in a questionnaire of the i-th industry; is the total number of preset questions in a questionnaire for the i-th industry, is the ranking weight of the jth key step in the ith industry; v is the total number of key step types corresponding to the questions in a questionnaire in the ith industry;

[0062] According to the evaluation objectives and process indicators of key steps, diagnostic rules are formulated and diagnostic expert evaluation logic is embedded.

[0063] The working principle of the above technical solution is as follows: First, according to the previously constructed indicator library, the key steps and process indicators are converted into specific questionnaire questions for various industry types, ensuring that the questionnaire questions can directly reflect the key business behaviors and process indicators of the enterprise. The design of questionnaire questions needs to consider the clarity, pertinence and answerability of the questions to ensure that the enterprise can answer them accurately and conveniently. According to the ranking weight of the key step type, the preliminary number of questionnaire questions corresponding to each key step in each industry is set.

[0064] Use the formula to calculate the number of questions that should correspond to each key step type; where Pn is the total number of questions in the preset questionnaire, and v is the total number of key step types involved in the questionnaire, ensuring that the number of questionnaire questions can reflect the importance of each key step in the business process, that is, the higher the ranking weight of the key step, the more questions it corresponds to. While designing the questionnaire questions, it is also necessary to formulate specific diagnostic rules based on the evaluation objectives and process indicators of the key steps. The diagnostic rules are the basis for analyzing and evaluating the answers to the questionnaire. It stipulates how to judge the business status, problems and improvement directions of the enterprise based on the answers of the enterprise. In addition, the evaluation logic of the diagnostic experts needs to be implanted into the questionnaire system. This includes the way experts interpret the answers to the questionnaire, the analysis process, and possible suggestions or improvement measures. The purpose of implanting the evaluation logic is to ensure that the output results of the questionnaire system can accurately reflect the real business status of the enterprise and provide valuable suggestions and improvement directions for the enterprise.

[0065] To sum up, the working principle of the S2 stage is mainly to convert the key steps and process indicators in the indicator library into questionnaire questions, set the number of questionnaire questions according to the ranking weights of the key steps, formulate diagnostic rules and implant evaluation logic, to build a questionnaire system that can accurately reflect the business status of the enterprise and provide improvement suggestions.

[0066] The effect of the above technical solution is: by directly converting key steps and process indicators into questionnaire questions, it ensures that the questionnaire content is closely related to the actual business of the enterprise, and improves the pertinence of the questionnaire. The number of questionnaire questions is allocated according to the ranking weights of key steps, ensuring that important steps receive more attention, thereby improving the effectiveness of the questionnaire. Setting the number of questionnaire questions according to the ranking weight avoids the situation where the number of questions is evenly distributed in the questionnaire, resulting in important steps being ignored, and optimizes resource allocation, so that limited questionnaire questions can more accurately cover the key links in the enterprise's business. Formulate diagnostic rules and implant diagnostic expert evaluation logic, so that the questionnaire system can conduct more in-depth analysis and evaluation based on the company's answers.

[0067] In some embodiments, the S3 includes:

[0068] Access to multiple diagnostic models;

[0069] Matching the output of the questionnaire assessment with the required input of the diagnostic model; obtaining a first matching degree;

[0070] The first matching degree is obtained by the following formula:

[0071] ;

[0072] Where u is the number of features common to the questionnaire output and the model input; ag is the weight of the g-th feature.

[0073] If the first matching degree between the output of the questionnaire evaluation and the demand input of multiple diagnostic models is greater than or equal to the first threshold, multiple candidate models will be selected in descending order according to the first matching degree;

[0074] If the first matching degree between the output of the questionnaire evaluation and the required input of each diagnostic model is less than the first threshold, the content of the questionnaire is adjusted according to the required input of the diagnostic model;

[0075] If the number of diagnostic models whose first matching degree between the output of the questionnaire evaluation and the required input of the diagnostic model is greater than or equal to the first threshold is less than the second threshold; then according to the number of the second threshold, in the order of the first matching degree from high to low, the corresponding candidate models are selected, and according to the input of the diagnostic model with a matching degree lower than the first, the questionnaire evaluation content is increased;

[0076] Through evaluation of alternative models, and based on historical evaluation results, the recognition of different models in various industries and cities is obtained, and based on the recognition, the preferred model is obtained.

[0077] The working principle of the above technical solution is as follows: First, multiple diagnostic models are collected and stored in advance. For example: the diagnostic model based on neural network, whose algorithm is based on neural network, especially deep learning technology, such as convolutional neural network (CNN), recurrent neural network (RNN), etc.; the diagnostic model based on fuzzy theory, whose theoretical basis is fuzzy theory, which allows processing uncertainty and ambiguity. By defining fuzzy sets and fuzzy rules, the model can process imprecise or ambiguous information. In diagnosis, this helps to identify problems that are difficult to describe with precise numbers. It is suitable for dealing with business problems with ambiguity, such as evaluating customer satisfaction and analyzing market trends; the diagnostic model based on Bayesian network, whose algorithm is based on Bayesian theorem, provides a method for calculating conditional probability. Complex probabilistic relationships are represented by constructing nodes (representing variables) and edges (representing the relationship between variables); in diagnosis, the model can infer unknown information based on known information; it is used to deal with business problems with uncertainty, such as fault diagnosis, risk assessment, etc. Rule-based diagnostic model; theoretical basis: expert system, which relies on the knowledge and experience of experts in a specific field. Working principle The model simulates the decision-making process of experts by defining a series of rules. These rules can be based on business logic, industry standards or expert experience. Applicable to business problems that require expert knowledge or experience, such as compliance checks, business process optimization, etc.; case-based diagnostic models (CBR); these models are based on different algorithms, theories or business logics, and are designed to provide enterprises with diagnostic services from different angles or levels. Next, the output of the questionnaire evaluation (i.e., the data answered by the enterprise) is matched with the demand input of each diagnostic model. This step is to determine which diagnostic models can process or analyze the data output by the questionnaire evaluation. The matching degree (first matching degree) can be obtained by calculating the similarity, correlation or conformity between the output of the questionnaire evaluation and the demand input of the diagnostic model. If the first matching degree between the output of the questionnaire evaluation and the demand input of multiple diagnostic models is greater than or equal to the preset first threshold, these models are regarded as candidate models. Select multiple candidate models in order from high to low according to the first matching degree for subsequent evaluation and comparison. If the first matching degree between the output of the questionnaire evaluation and the demand input of each diagnostic model is less than the first threshold, it means that the content of the current questionnaire may not be sufficient to meet the needs of any diagnostic model. In this case, the content of the questionnaire is adjusted according to the demand input of the diagnostic model to add questions or information related to the diagnostic model. If the number of diagnostic models whose first matching degree between the output of the questionnaire evaluation and the required input of the diagnostic model is greater than or equal to the first threshold is less than the preset second threshold, it means that although some models can process the questionnaire data, the number is insufficient. In this case, the corresponding candidate models are selected in descending order of the first matching degree, and the questionnaire evaluation content is increased according to the input of the diagnostic model with a lower matching degree than the first matching degree, in an attempt to improve the matching degree or increase the number of candidate models.Through the evaluation of the selected candidate models, the recognition of different models in various industries and cities is analyzed based on historical evaluation results; the recognition can be obtained by calculating the frequency of use, praise rate, satisfaction and other indicators of the model in a specific industry or city. Based on the recognition, the preferred models are further screened, that is, those with high recognition and good performance in various industries or cities; finally, the preferred models will be output for enterprises to choose and use. These preferred models can more accurately analyze the business data of enterprises and provide more valuable diagnostic results and improvement suggestions.

[0078] In the first matching degree, is an adjustment factor used to account for the difference in the number of features between the questionnaire and the model. When the number of features in the questionnaire and the model differs greatly, this adjustment factor will reduce the match to reflect this difference; if the number of features in the questionnaire and the model is exactly the same, the adjustment factor is 1 and does not affect the match.

[0079] The effect of the above technical solution is: by introducing feature weights and similarity calculations, the calculation formula of the first matching degree more accurately reflects the matching degree between the questionnaire evaluation output and the diagnostic model demand input. The introduction of the adjustment factor (min function part) further considers the difference in the number of questionnaire and model features, and improves the accuracy and rationality of the matching. The candidate models are selected in order from high to low according to the first matching degree, ensuring the objectivity and fairness of the selection process. When the matching degree is lower than the first threshold, the questionnaire content is adjusted in time to meet the model requirements, avoiding the model mismatch problem caused by improper questionnaire design. When the number of models with a matching degree higher than the first threshold is insufficient, the alternative model is selected according to the second threshold, and the questionnaire evaluation content is increased to improve the matching degree, ensuring the flexibility and practicality of model selection. Evaluation through alternative models reduces unnecessary model attempts and evaluation times, and improves evaluation efficiency. According to the historical evaluation results, the recognition of different models in various industries in various cities is obtained, which provides a reliable basis for the selection of the preferred model and avoids blind selection and waste of resources. The selection of the preferred model is based on historical evaluation results and recognition, ensuring the reliability and accuracy of the evaluation results. By continuously optimizing the questionnaire content and diagnostic model, the evaluation results can more truly reflect the actual situation and potential problems of the enterprise.

[0080] In some embodiments, the S4 includes:

[0081] Preprocessing the data collected from the questionnaire, including determining the completeness and validity of the questionnaire, eliminating invalid or abnormal data, formatting and coding;

[0082] The preprocessed data is input into the selected diagnostic model to obtain the diagnostic results.

[0083] The integrity and validity of the questionnaire include:

[0084] If the ratio of the number of missing questions to the total number of questions exceeds a third threshold, the questionnaire is judged to be incomplete and marked;

[0085] If the ratio of the number of missing questions to the total number of questions is less than or equal to the third threshold, but the missing ratio of questions corresponding to the key steps whose ranking weights are greater than the weight threshold is greater than the fourth threshold, the questionnaire is judged to be incomplete and marked;

[0086] Obtain answers to related questions in the questionnaire and perform semantic analysis. If there are logically inconsistent answers to the related questions, the answers to the related questions will be deemed invalid.

[0087] If the ranking weight of the key steps corresponding to the invalid questions is greater than the weight threshold, the questionnaire is marked invalid;

[0088] If the ranking weight of the key step corresponding to the invalid question is less than or equal to the weight threshold, but the ratio of the number of invalid questions to the total number of questions exceeds the fifth threshold, the questionnaire is marked invalid;

[0089] Questionnaires with missing marks were resurveyed, and questionnaires with invalid marks were excluded from the data analysis sample.

[0090] The working principle of the above technical solution is: the data collected by the questionnaire is input into the selected diagnostic model after a series of preprocessing steps to obtain accurate diagnostic results, specifically:

[0091] First, the ratio of the number of missing questions to the total number of questions in the questionnaire is calculated. If the ratio exceeds the preset third threshold, the questionnaire is determined to be incomplete and marked. In addition, even if the ratio of missing questions does not exceed the third threshold, if the missing ratio of questions corresponding to the key steps (determined according to the preset weight threshold) exceeds the fourth threshold, the questionnaire is also determined to be incomplete and marked.

[0092] Perform semantic analysis on the answers to related questions in the questionnaire to check whether there are any logically inconsistent answers; if logical inconsistency is found, the answers to the related questions are determined to be invalid; further, if the key step ranking weight corresponding to the invalid question is greater than the weight threshold, the questionnaire is marked invalid. Even if the key step ranking weight corresponding to the invalid question is not large, if the ratio of the number of invalid questions to the total number of questions exceeds the fifth threshold, the questionnaire is also marked invalid.

[0093] Based on the above judgment, questionnaire data marked as incomplete or invalid were eliminated.

[0094] The remaining questionnaire data are formatted, such as unifying the data format, converting the data type, etc., and necessary coding is performed to facilitate subsequent data analysis and model input.

[0095] The pre-processed valid questionnaire data is input into the previously selected diagnostic model; these models are built based on machine learning, deep learning or other statistical methods to analyze the questionnaire data and output diagnostic results.

[0096] The diagnostic results include an assessment of the company's current status, identification of potential problems, suggestions for improvement, etc. Questionnaires marked as missing require re-survey to obtain complete data; questionnaires marked as invalid are completely excluded from the data analysis sample to ensure the accuracy of subsequent analysis and diagnosis.

[0097] The above ensures that the data input into the diagnostic model is complete, valid and properly pre-processed, which helps to improve the accuracy and reliability of the diagnostic results.

[0098] The effect of the above technical solution is: by judging the completeness and validity of the questionnaire, missing or invalid questionnaires can be identified and marked, thereby ensuring that the data input into the diagnostic model is complete, accurate and reliable. Setting the third, fourth and fifth thresholds, as well as the sorting weights and weight thresholds of the key steps, provides clear quantitative standards for judging the completeness and validity of the questionnaire, and enhances the objectivity and accuracy of the judgment. The operations of eliminating invalid or abnormal data, formatting and encoding in the preprocessing step help to eliminate noise and inconsistency in the data and improve the quality and consistency of the data. Inputting high-quality data into the selected diagnostic model can more accurately analyze the questionnaire data and output more accurate and reliable diagnostic results. Re-surveying the questionnaires marked as missing can ensure the integrity of the data and avoid diagnostic result deviations caused by missing data. Excluding questionnaires marked as invalid from the data analysis sample can reduce unnecessary data processing and analysis work and improve resource utilization efficiency.

[0099] In some embodiments, the S5 includes:

[0100] Combine pictures, charts and animation elements to visualize the diagnostic report or interpretation content.

[0101] The principle and effect of the above technical solution is: by presenting complex data and information with intuitive and easy-to-understand visual elements (such as pictures, charts and animations), it can help users understand, analyze and remember more quickly.

[0102] In diagnostic reports or interpretations, using pictures, charts, and animations can more effectively convey key information, reveal relationships and trends between data, and demonstrate complex concepts and processes. These visual elements can enhance the readability and appeal of information, making reports more vivid, interesting, and easy to understand.

[0103] The present application provides an AI-based enterprise diagnostic report generation system, the system comprising:

[0104] The indicator library construction module is used to divide different industries according to business background and business model, divide the key steps of the business process of each industry through the funnel model and set process indicators, use cluster analysis and database statistics to obtain the ranking weights of different types of key steps, and then rank the process indicators in importance, add classification dimensions, and build a structured indicator library;

[0105] A questionnaire setting module is used to set evaluation questionnaires for different industries and the number of questions for different types of key steps through the indicator library, and to implant diagnostic expert evaluation logic;

[0106] The model selection module is used to set up multiple diagnostic models and select the appropriate diagnostic model based on the matching degree between the evaluation questionnaire content and the model and the recognition of the model by different cities;

[0107] A diagnosis module, used to collect enterprise data through an evaluation questionnaire, input the enterprise data into a selected diagnosis model, and obtain a diagnosis result;

[0108] Report generation module, used to generate diagnostic reports through AI.

[0109] In some embodiments, the system further includes an interaction module, wherein the interaction module includes:

[0110] A user information analysis unit, used to obtain user voice question information and analyze it;

[0111] The interactive output unit is used to generate a diagnosis report and / or interpret the diagnosis report based on the analysis results; and interpret the report in voice and / or text form.

[0112] In some embodiments, the indicator library construction module includes:

[0113] The first division unit is used to divide different industries according to business background and business model, where different industries include e-commerce, finance, education, and medical care;

[0114] The key step acquisition unit is used to acquire the business processes of various industries. According to the business logic and funnel model, each business process of each industry is divided into multiple key steps, and the process indicators of each step are acquired. The indicators can accurately reflect the key behaviors and conversion rates in the business process. For example, in the e-commerce industry, the business funnel includes steps such as user browsing, adding to cart, and purchasing. Corresponding process indicators can be set for each step.

[0115] The ranking weight acquisition unit is used to perform cluster analysis on each key step in the same industry to obtain the categories of the key steps; according to the number of times each key step appears in the database of the company in the industry and the first weight of the key step in each company, the ranking weight of each key step is obtained;

[0116] The ranking weight of the key steps is obtained by the following formula:

[0117] ;

[0118] in, is the ranking weight of the jth key step in the ith industry; is the first weight of the j-th key step in the ith industry among all the key steps of the k-th company; n is the number of times the j-th key step in the ith industry appears in different companies; m is the total category of the key steps in the ith industry;

[0119] An indicator ranking unit is used to set the importance ranking of the process indicators of the corresponding steps according to the ranking weights of each key step;

[0120] A dimension adding unit is used to add classification dimensions to process indicators of various key steps in various industries, wherein the classification dimensions include time dimensions and region dimensions;

[0121] The indicator library construction unit is used to build a structured indicator library through the database; and assign a unique code and name to each one.

[0122] In some embodiments, the questionnaire setting module includes:

[0123] The question quantity setting unit is used to convert key steps and process indicators into questionnaire questions for various types of industries according to the indicator library; according to the sorting weight of the key step type, the preliminary number of questionnaire questions corresponding to various types of key steps in various industries is set;

[0124] The preliminary number of key step questions is:

[0125] ;

[0126] in, is the number of questions corresponding to the jth key step in a questionnaire of the i-th industry; is the total number of preset questions in a questionnaire for the i-th industry, is the ranking weight of the jth key step in the ith industry; v is the total number of key step types corresponding to the questions in a questionnaire in the ith industry;

[0127] The diagnosis setting unit is used to formulate diagnosis rules and embed the diagnosis expert evaluation logic according to the evaluation objectives and process indicators of key steps.

[0128] In some embodiments, the model selection module includes:

[0129] A model acquisition unit, used for acquiring a variety of diagnostic models;

[0130] A first matching degree obtaining unit is used to match the output of the questionnaire evaluation with the required input of the diagnosis model to obtain a first matching degree;

[0131] The first matching degree is obtained by the following formula:

[0132] ;

[0133] Where u is the number of features common to the questionnaire output and the model input; ag is the weight of the g-th feature.

[0134] A first selection unit, configured to select a plurality of candidate models in descending order of the first matching degrees if the first matching degrees of the output of the questionnaire evaluation and the requirement inputs of the plurality of diagnostic models are both greater than or equal to a first threshold;

[0135] A first adjustment unit, configured to adjust the content of the questionnaire according to the requirement input of the diagnostic model if the first matching degree between the output of the questionnaire evaluation and the requirement input of each diagnostic model is less than a first threshold;

[0136] A second selection unit is used to select corresponding candidate models according to the number of the second threshold and in descending order of the first matching degree if the number of diagnostic models whose first matching degree between the output of the questionnaire evaluation and the required input of the diagnostic model is greater than or equal to the first threshold is less than the second threshold, and to increase the questionnaire evaluation content according to the input of the diagnostic model whose matching degree is lower than the first matching degree;

[0137] The third selection unit is used to evaluate through alternative models, and obtain the recognition of different models in various industries in various cities based on historical evaluation results, and obtain the preferred model based on the recognition.

[0138] In some embodiments, the diagnostic module includes:

[0139] A data preprocessing unit, used to preprocess the data collected by the questionnaire, said preprocessing including judging the integrity and validity of the questionnaire, eliminating invalid or abnormal data, formatting and coding;

[0140] The diagnosis unit is used to input the preprocessed data into the selected diagnosis model to obtain the diagnosis result.

[0141] In some embodiments, the data preprocessing unit includes:

[0142] A first marking unit, configured to determine that the questionnaire is incomplete and mark it if the ratio of the number of missing questions to the total number of questions exceeds a third threshold;

[0143] A second marking unit is used to judge that the questionnaire is incomplete and mark it if the ratio of the number of missing questions to the total number of questions is less than or equal to the third threshold, but the missing ratio of questions corresponding to the key steps whose ranking weights are greater than the weight threshold is greater than the fourth threshold;

[0144] The first judgment unit is used to obtain answers to related questions in the questionnaire and perform semantic analysis. If the answers to the related questions are logically inconsistent, the answers to the related questions are judged to be invalid.

[0145] A third marking unit is used to mark the questionnaire invalid if the ranking weight of the key step corresponding to the invalid question is greater than the weight threshold;

[0146] a fourth marking unit, for marking the questionnaire invalid if the ranking weight of the key step corresponding to the invalid question is less than or equal to the weight threshold, but the ratio of the number of invalid questions to the total number of questions exceeds a fifth threshold;

[0147] The sample processing unit is used to re-survey the questionnaires with missing marks and exclude the questionnaires with invalid marks from the data analysis sample.

[0148] In some embodiments, the report generation module includes: combining pictures, charts and animation elements to visually present the diagnostic report or interpretation content.

[0149] The working principle and effect of the above technical solution are the same as those in the embodiment of the method of the present application and will not be described in detail here.

[0150] The present application also provides an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of any method described in the present application are implemented.

[0151] This application is explained from the perspectives of purpose of use, effectiveness, progress and novelty, and has met the functional enhancement and usage requirements emphasized by the Patent Law. The above description and drawings of this application are only the preferred embodiments of this application, and are not intended to limit this application. Therefore, all structures, devices, features, etc. that are similar or identical to this application, that is, all equivalent replacements or modifications made in accordance with the scope of the patent application of this application, should fall within the scope of protection of the patent application of this application.

Claims

1. A method for generating enterprise diagnostic reports based on AI, characterized in that: The method comprises: S1. Different industries are divided according to business background and business model. The key steps of business processes in each industry are divided through the funnel model and process indicators are set. The ranking weights of different types of key steps are obtained by cluster analysis and database statistics. The importance of process indicators is ranked by ranking weights, and classification dimensions are added to build a structured indicator library. S2. Using the indicator library, set evaluation questionnaires for different industries and the number of questions for different types of key steps, and embed diagnostic expert evaluation logic; S3. Set up multiple diagnostic models, and select a suitable diagnostic model as the selected diagnostic model based on the matching degree between the content of the evaluation questionnaire and the model and the recognition of the model by different cities; S4. Collect enterprise data through the evaluation questionnaire, input the enterprise data into the selected diagnosis model, and obtain the diagnosis result; S5. Generate diagnostic reports through AI; The S3 includes: Access to multiple diagnostic models; Matching the output of the questionnaire assessment with the required input of the diagnostic model; obtaining a first matching degree; The first matching degree is obtained by the following formula: ； Where u is the number of features common to the questionnaire output and the model input; ag is the weight of the g-th feature; If the first matching degree between the output of the questionnaire evaluation and the demand input of multiple diagnostic models is greater than or equal to the first threshold, multiple candidate models will be selected in descending order according to the first matching degree; If the first matching degree between the output of the questionnaire evaluation and the required input of each diagnostic model is less than the first threshold, the content of the questionnaire is adjusted according to the required input of the diagnostic model; If the number of diagnostic models whose first matching degree between the output of the questionnaire evaluation and the required input of the diagnostic model is greater than or equal to the first threshold is less than the second threshold; then according to the number of the second threshold, in the order of the first matching degree from high to low, the corresponding candidate models are selected, and according to the input of the diagnostic model with a matching degree lower than the first, the questionnaire evaluation content is increased; Through the evaluation of alternative models, and based on the historical evaluation results, the recognition of different models in various industries and cities is obtained, and based on the recognition, the selected diagnostic model is obtained.

2. The method for generating an enterprise diagnostic report based on AI according to claim 1, characterized in that: The method further comprises: Obtain user voice question information and analyze it; Generate a diagnostic report and / or interpret the diagnostic report based on the analysis results; interpret the report in voice and / or text form.

3. The method for generating an enterprise diagnostic report based on AI according to claim 1, characterized in that: The S1 includes: Different industries are divided according to business background and business model. Obtain the business processes of various industries, divide the business processes of various industries into multiple key steps according to business logic and funnel model, and obtain process indicators for each step; For the same industry, cluster analysis is performed on each key step to obtain the categories of key steps; the ranking weights of each key step are obtained based on the number of times each key step appears in the database of companies in the industry and the first weight of the key step in each company; The ranking weight of the key steps is obtained by the following formula: ； in, is the ranking weight of the jth key step in the ith industry; is the first weight of the j-th key step in the ith industry among all the key steps of the k-th company; n is the number of times the j-th key step in the ith industry appears in different companies; m is the total category of the key steps in the ith industry; According to the ranking weight of each key step, set the importance ranking of the process indicators of the corresponding step; Adding classification dimensions to process indicators of various key steps in various industries, including time dimension and region dimension; Build a structured indicator library through the database; and assign a unique code and name to each one.

4. The method for generating an enterprise diagnostic report based on AI according to claim 1, characterized in that: The S2 includes: For each type of industry, the key steps and process indicators are converted into questionnaire questions based on the indicator library; according to the ranking weight of the key step type, the number of questionnaire questions corresponding to each type of key step in each industry is set; According to the evaluation objectives and process indicators of key steps, diagnostic rules are formulated and diagnostic expert evaluation logic is embedded.

5. The method for generating an enterprise diagnostic report based on AI according to claim 1, characterized in that: The S4 includes: Preprocessing the data collected from the questionnaire, including determining the completeness and validity of the questionnaire, eliminating invalid or abnormal data, formatting and coding; The preprocessed data is input into the selected diagnostic model to obtain the diagnostic results.

6. The method for generating an enterprise diagnostic report based on AI according to claim 5, characterized in that: The integrity and validity of the questionnaire include: If the ratio of the number of missing questions to the total number of questions exceeds a third threshold, the questionnaire is considered incomplete and marked; If the ratio of the number of missing questions to the total number of questions is less than or equal to the third threshold, but the missing ratio of questions corresponding to the key steps whose ranking weights are greater than the weight threshold is greater than the fourth threshold, the questionnaire is judged to be incomplete and marked; Obtain answers to related questions in the questionnaire and perform semantic analysis. If there are logically inconsistent answers to the related questions, the answers to the related questions will be deemed invalid. If the ranking weight of the key steps corresponding to the invalid questions is greater than the weight threshold, the questionnaire is marked invalid; If the ranking weight of the key step corresponding to the invalid question is less than or equal to the weight threshold, but the ratio of the number of invalid questions to the total number of questions exceeds the fifth threshold, the questionnaire is marked invalid; Questionnaires with missing marks were resurveyed, and questionnaires with invalid marks were excluded from the data analysis sample.

7. The method for generating an enterprise diagnostic report based on AI according to claim 1, characterized in that: The S5 includes: Combine pictures, charts and animation elements to visualize the diagnostic report or interpretation content.

8. An AI-based enterprise diagnostic report generation system, characterized in that: The system comprises: The indicator library construction module is used to divide different industries according to business background and business model, divide the key steps of the business process of each industry through the funnel model and set process indicators, use cluster analysis and database statistics to obtain the ranking weights of different types of key steps, and then rank the process indicators in importance, add classification dimensions, and build a structured indicator library; A questionnaire setting module is used to set evaluation questionnaires for different industries and the number of questions for different types of key steps through the indicator library, and to implant diagnostic expert evaluation logic; The model selection module is used to set up multiple diagnostic models and select the appropriate diagnostic model as the selected diagnostic model based on the content of the evaluation questionnaire and the matching degree of the model and the recognition of the model by different cities; A diagnosis module, used to collect enterprise data through an evaluation questionnaire, input the enterprise data into a selected diagnosis model, and obtain a diagnosis result; Report generation module, used to generate diagnostic reports through AI; The model selection module includes: Access to multiple diagnostic models; Matching the output of the questionnaire assessment with the required input of the diagnostic model; obtaining a first matching degree; The first matching degree is obtained by the following formula: ； Where u is the number of features common to the questionnaire output and the model input; ag is the weight of the g-th feature; If the first matching degree between the output of the questionnaire evaluation and the demand input of multiple diagnostic models is greater than or equal to the first threshold, multiple candidate models will be selected in descending order according to the first matching degree; If the first matching degree between the output of the questionnaire evaluation and the required input of each diagnostic model is less than the first threshold, the content of the questionnaire is adjusted according to the required input of the diagnostic model; If the number of diagnostic models whose first matching degree between the output of the questionnaire evaluation and the required input of the diagnostic model is greater than or equal to the first threshold is less than the second threshold; then according to the number of the second threshold, in the order of the first matching degree from high to low, the corresponding candidate models are selected, and according to the input of the diagnostic model with a matching degree lower than the first, the questionnaire evaluation content is increased; Through the evaluation of alternative models, and based on the historical evaluation results, the recognition of different models in various industries and cities is obtained, and based on the recognition, the selected diagnostic model is obtained.

9. An electronic device, characterized in that: It includes a memory, a processor, and a computer program stored in the memory and executable on the memory, wherein the processor executes the program to implement an AI-based enterprise diagnostic report generation method as described in any one of claims 1 to 7.

Images