Enterprise data mining system and method based on generative AI large model

Abstract

The invention relates to the technical field of enterprise data mining, in particular to an enterprise data mining system and method based on a generative AI large model, and the system comprises a problem improvement subsystem and a data mining subsystem. The data mining subsystem comprises a data calling module used for obtaining a corresponding structured data set from a database of the problem improvement subsystem; the model construction module is used for constructing an enterprise data mining model based on the AI large model; the model training module is used for training the constructed enterprise data mining model according to the called structured data set; the interaction module is used for acquiring technical question information; and based on the trained enterprise data mining model, processing the technical question information to generate technical feedback information, forming an enterprise data mining report corresponding to the technical question information according to a preset feedback template, and outputting the enterprise data mining report to the user.

Description

Technical Field

[0001] The present invention relates to the technical field of enterprise data mining, and specifically relates to an enterprise data mining system and method based on a generative AI large model. Background Art

[0002] With the development of information technology and the continuous improvement of enterprise informatization, enterprises and organizations have accumulated a large amount of business data. Rich information is hidden in these data, which can provide important support for enterprise decision-making, market prediction, customer service, etc.

[0003] However, traditional data analysis methods often rely on manual feature extraction. This method is not only time-consuming and laborious, but also difficult to discover deep-level rules and patterns from massive data.

[0004] In addition, traditional systems lack sufficient flexibility and adaptability when facing a rapidly changing data environment and are difficult to respond to the dynamic needs of enterprises in real time.

[0005] Based on this, there is an urgent need for an enterprise data mining system and method based on a generative AI large model, which can realize the functions of enterprise data mining automation and interactive feedback intelligence, and greatly improve the reliability and authenticity of enterprise data mining. Summary of the Invention

[0006] One of the purposes of the present invention is to provide an enterprise data mining system and method based on a generative AI large model.

[0007] To achieve the above purpose, an enterprise data mining system based on a generative AI large model is provided, which includes a problem improvement subsystem and a data mining subsystem;

[0008] The data mining subsystem includes:

[0009] A data retrieval module for retrieving a corresponding structured data set from the database of the problem improvement subsystem;

[0010] A model construction module for constructing an enterprise data mining model based on an AI large model;

[0011] A model training module for training the constructed enterprise data mining model according to the retrieved structured data set;

[0012] An interaction module for obtaining technical question information; and processing the technical question information based on the trained enterprise data mining model to generate technical feedback information, and forming an enterprise data mining report corresponding to the technical question information according to a preset feedback template and outputting it to the user.

[0013] Technical principle and effect of this solution: In this solution, the problem improvement subsystem is responsible for collecting, organizing, and storing structured data sets related to enterprise technical problems, providing basic data support for subsequent data mining.

[0014] Then, the data mining subsystem obtains the corresponding structured data set from the problem improvement subsystem to ensure that the data used for model training is accurate, complete, and representative, laying a solid foundation for subsequent model construction and training.

[0015] After that, an enterprise data mining model is constructed based on the AI large model, and the constructed enterprise data mining model is trained in combination with the retrieved structured data set. The model is trained with a large amount of historical data to enable it to accurately identify and understand the characteristics of technical problems, thereby improving the accuracy of prediction and analysis.

[0016] Finally, generative enterprise data mining can be achieved through the interaction module. Specifically, by obtaining the technical question information of the user, combining it with the trained enterprise data mining model, processing the technical question information, generating technical feedback information, and forming an enterprise data mining report corresponding to the question information according to the preset feedback template, and outputting it to the user.

[0017] In this solution, through automated data mining and intelligent analysis, the system can accurately identify the essence of the problem within a short time, quickly generate solutions, and significantly shorten the problem-solving cycle. Through the structured data set and the enterprise data mining model, the training of the enterprise data mining model based on the structured data set is realized, greatly improving the accuracy, timeliness, and feasibility of model answering and data mining according to the user's technical question information, that is, realizing the functions of enterprise data mining automation and interactive feedback intelligence, and greatly improving the reliability and authenticity of enterprise data mining.

[0018] The interaction module realizes a human-computer dialogue interaction experience. Users can describe problems in natural language, obtain instant technical feedback and detailed solution reports, which greatly facilitates the use of users.

[0019] Furthermore, the problem improvement subsystem includes a question module, a cause analysis module, a measure determination module, and an evaluation feedback module;

[0020] The question module is used for users to publish problem data to the enterprise platform and also for pushing the problem data to the corresponding person in charge for processing;

[0021] The cause analysis module is used to analyze the corresponding cause when the user receives the problem data, form cause data, publish the cause data to the enterprise platform and associate it with the problem data;

[0022] The measure determination module is used to formulate corresponding improvement measure data according to the cause data, and publish the improvement measure data to the enterprise platform and associate it with the cause data and the problem data;

[0023] The evaluation feedback module is used to allow users to evaluate the process of problem handling to form evaluation data;

[0024] The storage module is used to store the problem data, cause data, improvement measure data and evaluation data into the corresponding database to form a structured data set.

[0025] Beneficial effects: In this solution, by allowing users to publish problem data to the enterprise platform, the sharing of problem data corresponding to the entire enterprise is realized, and the problem data will also be pushed to the corresponding person in charge. Through this push of problem data, it is ensured that the problem data can be received and processed by relevant personnel in a timely manner, shortening the problem-solving cycle while realizing the sharing of problem data. Each user within the enterprise can not only view the problem data on the enterprise platform, but also quickly solve the problem data to be processed by themselves.

[0026] After the corresponding person in charge receives the problem data, they will analyze the cause corresponding to the problem data and associate and publish the corresponding cause data with the problem data. Through this step, the cause analysis of the problem data and the sharing of the cause are realized. Associating the cause data with the problem data enhances the transparency of the problem handling process and facilitates traceability and review.

[0027] Then, after determining the corresponding cause data, the corresponding improvement measure data will be formulated, and the improvement measure data will be published to the enterprise platform and associated with the cause data and the problem data. By sequentially publishing and associating the problem data, cause data, and improvement measure data to the enterprise platform, it is ensured that all relevant personnel within the enterprise platform can view the complete processing process on a unified platform. That is, the publication of the improvement measure data enables all relevant parties (such as technicians, managers, users, etc.) to obtain the latest processing progress in a timely manner, reducing delays and misunderstandings in information transmission. Users and management can clearly understand the ins and outs of each problem, including the origin of the problem, the analyzed cause, and the final improvement measures taken.

[0028] The associated data forms a valuable knowledge base, recording the solutions and improvement measures for various problems, providing a reference basis for future similar problems. New employees or external partners can quickly learn and draw on past successful experiences through these associated data, improving the overall work efficiency.

[0029] A corresponding structured data set is constituted by problem data, cause-of-generation data, improvement measure data, and evaluation data. When using the structured data for training an enterprise data mining model subsequently, since the corresponding structured data set contains the complete process information from problem raising to solution, the model can understand the essence and processing process of the problem more comprehensively, better understand the causal relationship between the cause of the problem and the solution, enhance the interpretability and transparency of the model, and the diverse data types provide rich features, which helps the model capture more complex relationships and patterns, thereby improving the accuracy of the model in prediction and classification.

[0030] Each problem processed and its corresponding solution are recorded, forming a valuable knowledge base, which not only helps solve the current problem but also provides a reference for future similar problems, that is, the data required for constructing the corresponding model is currently occurring in real time and has been solved in real time. Compared with the existing model construction that requires collecting historical problems, annotating and processing historical data, the data corresponding to the model of this application is more real and real-time, realizing the automatic generation of the structured data set, and greatly improving the reliability and speed of model training.

[0031] Furthermore, the interaction module is used to obtain the technical question information corresponding to the user; according to the technical question information corresponding to the user, determine the interaction level corresponding to the user, and based on the interaction level, determine the corresponding question type, forming a set of question types corresponding to the user;

[0032] It is also used to match the question type corresponding to the technical question information of the user based on the formed set of question types, and judge whether the question type corresponding to the technical question information is in the set of question types. If not, a reminder message is fed back to the user. Otherwise, according to the question type corresponding to the technical question information, the corresponding enterprise data mining model is retrieved from the database, the technical question information is used as input data and input into the corresponding enterprise data mining model, the corresponding technical feedback information is output, and based on a preset feedback template, the technical feedback information is integrated into the corresponding enterprise data mining report and fed back to the user;

[0033] It is also used to count the number of times the judgment result is no, form the corresponding unauthorized access times, and based on the preset question restriction strategy, restrict the acquisition of the technical question information corresponding to the user based on the corresponding unauthorized access times.

[0034] Beneficial effects: In this solution, after obtaining the technical question information corresponding to the user, according to the technical question information of the user, the corresponding interaction level of the user is determined, and the corresponding question type is determined based on the interaction level, forming a set of question types corresponding to the user. By determining the corresponding set of question types, the question permission of the user can be quickly checked.

[0035] After that, it is judged whether the question type corresponding to the technical question information is in the set of question types. If not, a reminder message is fed back to the user; otherwise, according to the question type corresponding to the technical question information, the corresponding enterprise data mining model is retrieved from the database, the technical question information is used as input data and input into the corresponding enterprise data mining model, the corresponding technical feedback information is output, and based on a preset feedback template, the technical feedback information is integrated into the corresponding enterprise data mining report and fed back to the user. By judging whether the question type of the user's technical question information is in the set of question types, it is determined whether the user has the permission to execute the technical feedback information corresponding to the technical question information. Through this step of judgment, the user can immediately receive a confirmation or reminder message about the question type, ensuring that the problem can be correctly processed. Through strict question type matching, invalid or irrelevant questions are prevented from occupying system resources, ensuring that resources are used for users with real needs.

[0036] At the same time, during the process of the user uploading technical question information, the number of times when the question type is not in the set of question types is counted, and combined with the corresponding question restriction strategy, the corresponding question restriction is realized, preventing the user from frequently submitting invalid or irrelevant questions, avoiding the abuse of system resources. By restricting invalid questions, the proportion of valid questions on the platform is increased, enabling technical personnel to focus more on solving practical problems and improving the overall service quality. For normal users, a relaxed question environment is maintained; while for users who frequently have no permission to access, the restriction intensity is gradually increased, which not only maintains system security but also does not overly affect the user experience.

[0037] Further, the structured data set obtained from the database in the problem improvement subsystem includes the structured data set corresponding to a certain question type or all the structured data sets in the database.

[0038] Beneficial effects: In this solution, the obtained structured data set can be either the structured data set corresponding to a certain question type or all the structured data sets in the database, enabling subsequent training of the enterprise data mining model to train both a highly targeted model corresponding to a certain question type and a model with strong generality, greatly improving the adaptability of the corresponding enterprise data mining model.

[0039] Further, the preset question restriction strategy is:

[0040] When the number of unauthorized accesses corresponding to the user is less than or equal to the first preset number threshold, the acquisition of the technical question information corresponding to the user is not restricted;

[0041] If the number of unauthorized accesses corresponding to the user is greater than the first preset number threshold and less than or equal to the second preset number threshold, then within a preset time period, the number of acquisitions corresponding to the acquisition of the technical question information corresponding to the user does not exceed the acquisition number threshold, and when matching the question types corresponding to the technical question information of the user, a preset proportion of the question types in the question type set corresponding to the user is prohibited from being matched;

[0042] If the number of unauthorized accesses corresponding to the user is greater than the second preset number threshold, then within a preset time period, the acquisition of the technical question information corresponding to the user is prohibited.

[0043] Beneficial effects: In this solution, by setting different thresholds to restrict the user's questioning behavior, it can effectively prevent malicious users or automated programs from abusing system resources, thereby protecting the stability of the system and the security of data. Different levels of restriction measures are set in the policy, showing tolerance to users who make mistakes occasionally, while implementing stricter restrictions on users who frequently violate the regulations. This flexibility helps to safeguard the rights and interests of most legitimate users while effectively curbing potential threats. Appropriate restrictions on unauthorized access behavior exceeding a certain number of times can reduce unnecessary consumption of computing and storage resources, ensuring that limited resources are preferentially served for compliant query requests.

[0044] The present invention also provides an enterprise data mining method based on a generative AI large model, using the above-mentioned enterprise data mining system based on a generative AI large model. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a logical block diagram of the enterprise data mining system based on a generative AI large model in Embodiment 1 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0046] The following is a further detailed description through specific embodiments:

[0047] Embodiment 1

[0048] An enterprise data mining system based on a generative AI large model is basically as Figure 1 shown, including a problem improvement subsystem and a data mining subsystem;

[0049] The problem improvement subsystem includes a question module, a cause analysis module, a measure determination module, and an evaluation feedback module;

[0050] The question module is used for the user to post question data to the enterprise platform and also for pushing the question data to the corresponding person in charge for processing. In this embodiment, when people in the enterprise encounter problems in daily life, such as a certain part often having a certain problem, they can post the corresponding questions on the enterprise platform. At this time, according to the questions raised by this person, they are pushed to the corresponding person in charge, such as the maintenance personnel mainly responsible for maintenance.

[0051] The cause analysis module is used for analyzing the corresponding cause of generation when receiving the question data by the user, forming cause of generation data, posting the cause of generation data to the enterprise platform and associating it with the question data. In this embodiment, when the user receives the question data, they will analyze the question, thus informing the cause of generation and posting it to the enterprise platform to be associated with the corresponding question data. In the enterprise platform, each person in the enterprise can be both the publisher and the handler of the question. The multi-role playing makes each person in the enterprise more sticky to the enterprise platform and have better interactivity.

[0052] The measure determination module is used for formulating corresponding improvement measure data according to the cause of generation data, and posting the improvement measure data to the enterprise platform and associating it with the cause of generation data and the question data;

[0053] The evaluation feedback module is used for the user to evaluate the process of question handling and form evaluation data;

[0054] The storage module is used for storing the question data, the cause of generation data, the improvement measure data and the evaluation data into the corresponding database to form a structured data set. In this embodiment, the whole process from the question being raised to analyzed, then to solved and the subsequent solution evaluation is associated to form a structured data set, providing a reliable and solid data basis for subsequent model training.

[0055] The data mining subsystem includes:

[0056] The data retrieval module is used for obtaining the corresponding structured data set from the database of the question improvement subsystem; the structured data set obtained from the database in the question improvement subsystem includes the structured data set corresponding to a certain question type or all the structured data sets in the database. In this embodiment, the enterprise data mining models trained by different structured data sets are different, some are highly targeted and some are highly general, greatly improving the adaptability of the models.

[0057] The model construction module is used for constructing an enterprise data mining model based on the AI large model;

[0058] A model training module for training the constructed enterprise data mining model according to the retrieved structured data set;

[0059] An interaction module for obtaining technical question information; and based on the trained enterprise data mining model, processing the technical question information to generate technical feedback information, and forming an enterprise data mining report corresponding to the technical question information according to a preset feedback template and outputting it to the user.

[0060] The interaction module is used to obtain the technical question information corresponding to the user; determine the interaction level corresponding to the user according to the technical question information corresponding to the user, and determine the corresponding question type based on the interaction level to form a set of question types corresponding to the user;

[0061] It is also used to match the question type corresponding to the technical question information of the user based on the formed set of question types, and judge whether the question type corresponding to the technical question information is in the set of question types. If not, a reminder message is fed back to the user. Otherwise, according to the question type corresponding to the technical question information, the corresponding enterprise data mining model is retrieved from the database, the technical question information is used as input data and input into the corresponding enterprise data mining model, the corresponding technical feedback information is output, and based on the preset feedback template, the technical feedback information is integrated into the corresponding enterprise data mining report and fed back to the user;

[0062] It is also used to count the number of times when the judgment result is negative to form the corresponding unauthorized access times, and restrict the acquisition of the technical question information corresponding to the user based on the corresponding unauthorized access times according to the preset question restriction strategy.

[0063] The preset question restriction strategy is:

[0064] If the number of unauthorized access times corresponding to the user is less than or equal to the first preset number threshold, the acquisition of the technical question information corresponding to the user is not restricted;

[0065] If the number of unauthorized access times corresponding to the user is greater than the first preset number threshold and less than or equal to the second preset number threshold, within a preset time period, the number of acquisitions corresponding to the acquisition of the technical problem information corresponding to the user does not exceed the acquisition number threshold, and when matching the question type corresponding to the technical question information of the user, a preset proportion of the question types in the set of question types corresponding to the user are prohibited from being matched;

[0066] If the number of unauthorized accesses corresponding to the user is greater than the second preset number threshold, then within a preset time period, obtaining the technical problem information corresponding to the user is prohibited. In this embodiment, the preset time period is dynamically variable. It is not only different within two different thresholds, but also within the same threshold, the corresponding preset time period will increase with the difference between the number of unauthorized accesses and the corresponding threshold. For example, when the number of unauthorized accesses corresponding to the user is greater than the first preset number threshold and less than or equal to the second preset number threshold, the initial preset time period is A. As the number of unauthorized accesses of the user increases, for every increase of one-tenth of the difference between the second preset number threshold and the first preset number threshold, the corresponding preset time period increases by a. That is, the preset time period T = A + na, where n is the number of times of increasing one-tenth of the difference between the second preset number threshold and the first preset number threshold. And when the number of unauthorized accesses corresponding to the user is greater than the second preset number threshold, the initial preset time period is A + 10a. After that, for every increase of 10 times in the number of unauthorized accesses of the user, the corresponding preset time period increases by b. When the preset time period increases to the tenth time, the corresponding preset time period directly becomes 24 hours.

[0067] This embodiment also discloses an enterprise data mining method based on a generative AI large model, using the above-mentioned enterprise data mining system based on a generative AI large model.

[0068] The above are only embodiments of the present invention. Well-known specific structures and characteristics and other common knowledge are described too much in the solution. Those of ordinary skill in the art know all the common technical knowledge in the technical field to which the invention belongs before the application date or the priority date, can know all the existing technologies in this field, and have the ability to apply the conventional experimental means before this date. Those of ordinary skill in the art can, under the inspiration given in this application, combine their own abilities to complete and implement this solution. Some typical well-known structures or well-known methods should not become an obstacle for those of ordinary skill in the art to implement this application. It should be noted that for those skilled in the art, without departing from the structure of the present invention, several deformations and improvements can still be made, and these should also be regarded as the protection scope of the present invention, and these will not affect the implementation effect of the present invention and the practicality of the patent. The protection scope required by this application should be subject to the content of its claims, and the specific implementation manners and the like recorded in the specification can be used to explain the content of the claims.

Claims

1. An enterprise data mining system based on a generative AI big model, characterized by: Includes problem improvement subsystem and data mining subsystem; The data mining subsystem includes: The data retrieval module is used to obtain the corresponding structured data set from the database of the problem improvement subsystem; Model building module, used to build enterprise data mining models based on AI big models; Model training module, used to train the constructed enterprise data mining model based on the retrieved structured data set; The interactive module is used to obtain technical question information; and based on the trained enterprise data mining model, the technical question information is processed to generate technical feedback information, and the enterprise data mining report corresponding to the technical question information is formed according to the preset feedback template and output to the user.

2. The enterprise data mining system based on the generative AI big model according to claim 1 is characterized by: The problem improvement subsystem includes a questioning module, a cause analysis module, a measure determination module and an evaluation feedback module; The question module is used for users to publish question data on the enterprise platform and also for pushing question data to corresponding persons in charge for processing; The cause analysis module is used to analyze the corresponding cause of the problem data received by the user, generate cause data, publish the cause data to the enterprise platform and associate it with the problem data; The measure determination module is used to formulate corresponding improvement measure data according to the cause data, and publish the improvement measure data on the enterprise platform and associate it with the cause data and the problem data; The evaluation feedback module is used for users to evaluate the process of problem handling and form evaluation data; The storage module is used to store the problem data, cause data, improvement measure data and evaluation data into a corresponding database to form a structured data set.

3. The enterprise data mining system based on the generative AI big model according to claim 2 is characterized by: The interaction module is used to obtain the technical question information corresponding to the user; determine the interaction level corresponding to the user according to the technical question information corresponding to the user, and determine the corresponding question type based on the interaction level to form a question type set corresponding to the user; It is also used to match the question type corresponding to the technical question information corresponding to the user based on the formed question type set, and determine whether the question type corresponding to the technical question information is in the question type set. If not, a reminder message is fed back to the user. Otherwise, according to the question type corresponding to the technical question information, a corresponding enterprise data mining model is retrieved from the database, the technical question information is used as input data, and is input into the corresponding enterprise data mining model, and the corresponding technical feedback information is output. Based on a preset feedback template, the technical feedback information is integrated into a corresponding enterprise data mining report and fed back to the user; It is also used to count the number of times the judgment result is negative, form the corresponding number of unauthorized access, and according to the preset question restriction strategy, based on the corresponding number of unauthorized access, restrict the acquisition of the technical question information corresponding to the user.

4. The enterprise data mining system based on the generative AI big model according to claim 3 is characterized by: The structured data set obtained from the database in the question improvement subsystem includes a structured data set corresponding to a certain question type or all structured data sets in the database.

5. The enterprise data mining system and method based on the generative AI big model according to claim 4 is characterized by: The preset question restriction strategy is: If the number of unauthorized accesses corresponding to the user is less than or equal to the first preset number threshold, the acquisition of the technical question information corresponding to the user is not restricted; If the number of unauthorized accesses corresponding to the user is greater than the first preset number threshold and less than or equal to the second preset number threshold, then in the preset time period, the number of acquisitions corresponding to the acquisition of the technical question information corresponding to the user does not exceed the acquisition number threshold, and when matching the question types corresponding to the technical question information corresponding to the user, the question types corresponding to the user's question types that are in a preset proportion are prohibited from matching; If the number of unauthorized accesses corresponding to the user is greater than the second preset number threshold, then in the preset time period, it is prohibited to obtain the technical problem information corresponding to the user.

6. An enterprise data mining method based on a generative AI big model, characterized by: An enterprise data mining system based on a generative AI big model using any one of claims 1 to 5 above.

Images