AI learning method and system based on question supervision

Abstract

The invention relates to the technical field of model training, in particular to an AI learning method and system based on question supervision, and the method comprises the steps: obtaining the whole-process data of a question as a training data set, and training an AI model: obtaining the whole-process data of the question, taking all the whole-process data except evaluation data as an input, taking the evaluation data as an output, and taking the evaluation data as an output; positive evaluation and negative evaluation in the evaluation data are used as data annotations; the AI model is trained through a training data set formed by input and output, so that the AI model can analyze and evaluate problems; wherein the whole process data of the problem comprises all data generated in the whole process from initiation to solution of the problem and subsequent evaluation data, and the evaluation data comprises one or more of evaluation information of a problem initiator, evaluation information of a supervision leader and evaluation information of a public problem. According to the scheme, multidirectional supervision can be formed, the completion efficiency and the completion quality are guaranteed, and meanwhile the data utilization rate can be increased.

Description

Technical Field

[0001] The present invention relates to the technical field of model training, and specifically to an AI learning method and system based on issue supervision and handling. Background Art

[0002] Aiming at the problems of low efficiency in problem communication and task execution caused by the lack of a timely and effective communication mechanism, unclear task assignment, and lack of a timely and effective feedback mechanism. Researchers have proposed that instead of simply relying on system improvement, which will increase the workload and have little effect, it is to assist work through a set system, which is convenient to implement, will not increase the workload, has good efficiency, and optimizes problem improvement management to solve the problems of low efficiency in problem communication and task execution and improve work efficiency.

[0003] However, a large amount of data is generated during the process from the generation to the solution of the problem, but it is not effectively utilized. And after the problem is released, there is no progress supervision, and the completion efficiency and quality are not ideal.

[0004] Therefore, there is an urgent need for an AI learning method and system based on issue supervision and handling, which can form multi-faceted supervision, ensure the completion efficiency and quality, and at the same time improve the data utilization rate and effectively utilize the data generated during the process from the generation to the solution of the problem. Summary of the Invention

[0005] One of the purposes of the present invention is to provide an AI learning system based on issue supervision and handling, which can form multi-faceted supervision, ensure the completion efficiency and quality, and at the same time improve the data utilization rate and effectively utilize the data generated during the process from the generation to the solution of the problem.

[0006] The first basic solution provided by the present invention: An AI learning system based on issue supervision and handling, comprising:

[0007] A training module, configured to obtain the whole-process data of the problem as a training data set to train the AI model, including:

[0008] Obtain the whole-process data of the problem, use all the whole-process data except the evaluation data as the input, and the evaluation data as the output, and the positive evaluation and negative evaluation in the evaluation data are used as data annotations;

[0009] Train the AI model with the training data set formed by the input and output, so that the AI model can analyze and evaluate the problem;

[0010] Wherein the whole-process data of the problem includes all the data generated during the whole process from the initiation to the solution of the problem, and the subsequent evaluation data, and the evaluation data includes one or more of the evaluation information of the problem initiator, the evaluation information of the supervision leader, and the evaluation information of the publicly disclosed problem.

[0011] Beneficial effects of the first basic solution: This solution obtains the whole-process data of the problem as the training data set to train the AI model, thereby improving the data utilization rate. The trained AI model can analyze and evaluate the problem, which is beneficial to further data analysis. And the whole-process data of the problem includes all the data generated during the whole process from the initiation to the solution of the problem, as well as the subsequent evaluation data. The evaluation data includes one or more of the evaluation information of the problem initiator, the evaluation information of the supervisor leader, and the evaluation information of the publicly disclosed problem. Through the evaluation method, the problem initiator, the supervisor leader, and the masses can all pay attention to the problem. During the process of problem rectification, supervision by the leader, the problem initiator, and the masses in three aspects forms multi-faceted supervision, ensuring the completion efficiency and quality. The person in charge of the problem can complete the problem rectification process more efficiently and with high quality.

[0012] This solution can form multi-faceted supervision, ensure the completion efficiency and quality, and at the same time improve the data utilization rate, effectively utilizing the data generated from the problem occurrence to the solution.

[0013] Furthermore, the training module is used to obtain the whole-process data of the problem as the training data set to train the AI model, and further includes:

[0014] Extract part of the input and output to form a test set, use the test set to evaluate the performance of the trained model, and adjust the parameters or structure of the AI model according to the evaluation results. This is convenient for the adjustment and optimization of the AI model.

[0015] Furthermore, it further includes: an information acquisition module, a display module, a user module, and an evaluation module;

[0016] The information acquisition module is used to obtain the public selection information of the problem to be solved, the selection information of the problems under attention and supervision, and the evaluation information of the publicly disclosed problems;

[0017] The display module is used to obtain the problem to be solved for public disclosure according to the public selection information of the problem to be solved;

[0018] The user module is used to obtain the corresponding problem according to the selection information of the problems under attention and supervision and add it to the user's attention and supervision list;

[0019] The evaluation module is used to receive the evaluation information of the publicly disclosed problems and display it in association with the publicly disclosed problems. By the user module selecting the problems, paying attention to and supervising the problems, supervision in three aspects of the leader, the problem initiator, and the masses is formed. The person in charge of the problem can complete the problem rectification process more efficiently and with high quality, achieving the effects of key attention, full-staff supervision, and efficient handling.

[0020] Furthermore, it further includes: a permission management module;

[0021] A permission management module is used to set user permissions and issue permissions; users can view issues with corresponding issue permissions according to their user permissions. This facilitates hierarchical management of issues and confidentiality of relevant content.

[0022] The second object of the present invention is to provide an AI learning method based on issue supervision, which can form multi-faceted supervision, ensure completion efficiency and quality, and at the same time improve data utilization rate and effectively utilize the data generated during the process from issue generation to solution.

[0023] The present invention provides the second basic solution: an AI learning method based on issue supervision, including:

[0024] Obtain the whole-process data of the issue as the training data set to train the AI model, including:

[0025] Obtain the whole-process data of the issue. Use all the whole-process data except the evaluation data as the input, and the evaluation data as the output. And in the evaluation data, positive evaluations and negative evaluations are used as data annotations;

[0026] Train the AI model with the training data set formed by the input and output, so that the AI model can analyze and evaluate the issue.

[0027] Among them, the whole-process data of the issue includes: all the data generated during the whole process from the issue initiation to solution, and subsequent evaluation data. Among them, the evaluation data includes: one or more of the evaluation information of the issue initiator, the evaluation information of the supervisor, and the evaluation information of the publicly disclosed issue.

[0028] Furthermore, the step of obtaining the whole-process data of the issue as the training data set to train the AI model further includes:

[0029] Extract part of the input and output to form a test set, use the test set to evaluate the performance of the trained model, and adjust the parameters or structure of the AI model according to the evaluation results. This is convenient for the adjustment and optimization of the AI model.

[0030] Furthermore, it further includes:

[0031] Obtain the public selection information of the issue to be solved, the selection information of the issue concerned and supervised, and the evaluation information of the publicly disclosed issue;

[0032] According to the public selection information of the issue to be solved, make the issue to be solved public;

[0033] According to the selection information of the issue concerned and supervised, obtain the corresponding issue and add it to the user's concerned and supervised list;

[0034] Receive the evaluation information of the disclosed problems and display it in association with the disclosed problems. By selecting problems, pay attention to the supervised problems, form supervision by leaders, problem initiators, and the masses. The person in charge of the problem can complete the rectification process of the problem more efficiently and with higher quality, achieving the effects of key attention, full-staff supervision, and efficient handling.

[0035] Furthermore, it also includes: setting user permissions and problem permissions; users can view problems corresponding to the problem permissions according to their user permissions. This is convenient for hierarchical management of problems and the confidentiality of relevant content. Brief Description of the Drawings

[0036] Figure 1 It is a logic block diagram of an embodiment of an AI learning system based on problem supervision of the present invention. Detailed Embodiment

[0037] The following is a further detailed description through specific embodiments:

[0038] Embodiment 1

[0039] The embodiment is basically as shown in the attached Figure 1 : An AI learning system based on problem supervision includes:

[0040] A training module for obtaining the whole-process data of the problem as a training data set to train the AI model;

[0041] Specifically, obtain all the data generated during the whole process from the initiation to the solution of the problem, that is, the whole-process data of the problem, as the input; evaluation data as the output, and positive evaluations and negative evaluations in the evaluation data are used as data annotations for the AI model to learn; classify whether the evaluation data is a positive evaluation or a negative evaluation through data annotation, so that the subsequent AI model can distinguish whether the given evaluation is a positive evaluation or a negative evaluation after training;

[0042] The training data set formed by the input and output is used to train the AI model so that the AI model can analyze and evaluate the problem; the AI model extracts the features in the whole-process data of the problem and establishes the mapping relationship between the features and the evaluation data, so as to generate the corresponding evaluation data through the recognition of the features in the whole-process data of the problem next time; the AI model can adopt decision trees, random forests, support vector machines, neural networks, etc.;

[0043] In addition, part of the input and output will be extracted to form a test set, use the labeled data to train the model, and ensure good generalization ability of the model through cross-validation; evaluate the model performance on the independent test set, and adjust the model parameters or structure according to the results to optimize its performance; deploy the trained AI model to the actual environment, evaluate the problem handling effect of new cases in real time, and continuously collect data for model update.

[0044] Among them, the whole-process data of the problem includes: all data generated during the whole process from the initiation to the solution of the problem, as well as subsequent evaluation data. The evaluation data includes, but is not limited to: one or more of the evaluation information uploaded or directly obtained by the problem initiator through the terminal, the evaluation information uploaded or directly obtained by the supervisor leader through the terminal, and the evaluation information of the publicly received problems.

[0045] All data generated during the whole process from the initiation to the solution of the problem is generated by the following modules:

[0046] The information acquisition module is used to acquire the publicly selected information of the problem to be solved, the selected information of the problem under supervision and attention, and the evaluation information of the publicly received problems.

[0047] The display module is used to publicly display the problem to be solved according to the publicly selected information of the problem to be solved; all users can view the publicly displayed problems through the display module and input evaluation information through the information acquisition module.

[0048] The user module is used to acquire the corresponding problems according to the selected information of the problems under supervision and attention and add them to the user's supervision and attention list; among them, the users are mainly enterprise leaders who select the problems to be supervised and attended through the terminal.

[0049] The evaluation module is used to receive the evaluation information of the publicly received problems and display them in association with the publicly received problems; the evaluation module is also used to call the AI model for evaluation.

[0050] This solution forms a supervision by leaders, problem initiators and the masses. The problem responsible person can complete the rectification process of the problem more efficiently and with higher quality, achieving the effects of key attention, full-staff supervision and efficient handling.

[0051] Embodiment 2

[0052] This embodiment is basically the same as the above embodiment, except that: it further includes: a permission management module;

[0053] The permission management module is used to set user permissions and problem permissions; users can view the problems with corresponding problem permissions according to their user permissions.

[0054] Specifically, the permission management module is used to set user permissions and problem permissions. In this embodiment, user permissions and problem permissions are set according to the job ranks of enterprise employees. For example, the job ranks range from 1 to 5, and the larger the value, the higher the job rank. The corresponding user permissions and problem permissions also range from 1 to 5; users with a user permission of 4 can view the problems with problem permissions of 1-4, that is, their display module will only display the problems with problem permissions of 1-4 and their evaluation information, so as to facilitate the hierarchical management of problems and the confidentiality of relevant content.

[0055] Example 3

[0056] An AI learning method based on problem supervision and handling includes:

[0057] Obtain the publicly selected information of the problems to be solved, the selected information of the problems concerned with supervision, and the evaluation information of the publicly available problems;

[0058] According to the publicly selected information of the problems to be solved, obtain the problems to be solved for public disclosure;

[0059] According to the selected information of the problems concerned with supervision, obtain the corresponding problems and add them to the user's supervision and handling list; where the user is mainly an enterprise leader, and selects the problems to be supervised and handled through a terminal;

[0060] Receive the evaluation information of the publicly available problems and display them in association with the publicly available problems; an AI model can also be called for evaluation;

[0061] Obtain the whole-process data of the problems as the training data set to train the AI model;

[0062] Specifically, obtain all the data generated from the initiation to the solution of the problems, that is, the whole-process data of the problems, as the input; the evaluation data as the output, and the positive evaluations and negative evaluations in the evaluation data are used as data annotations for the AI model to learn; classify whether the evaluation data is a positive evaluation or a negative evaluation through data annotation, so that the subsequent AI model can distinguish whether the given evaluation is a positive evaluation or a negative evaluation after training;

[0063] The training data set formed by the input and output is used to train the AI model so that the AI model can analyze and evaluate the problems; the AI model extracts the features in the whole-process data of the problems and establishes the mapping relationship between the features and the evaluation data, so as to generate the corresponding evaluation data through the recognition of the features in the whole-process data of the problems next time; the AI model can adopt decision trees, random forests, support vector machines, neural networks, etc.;

[0064] In addition, part of the input and output will be extracted to form a test set, and the data with good labels will be used to train the model, and the generalization ability of the model will be ensured to be good through cross-validation; evaluate the model performance on an independent test set, and adjust the model parameters or structure according to the results to optimize its performance; deploy the trained AI model to the actual environment, evaluate the problem handling effect of new cases in real time, and continuously collect data for model update.

[0065] The whole-process data of the problem includes: all data generated during the whole process from the initiation to the solution of the problem, as well as subsequent evaluation data. The evaluation data includes, but is not limited to: one or more of the evaluation information uploaded or directly obtained by the problem initiator through the terminal, the evaluation information uploaded or directly obtained by the supervisor leader through the terminal, and the evaluation information of the publicly received problems.

[0066] This solution forms supervision from three aspects: leaders, problem initiators, and the masses. The person in charge of the problem can complete the rectification process of the problem more efficiently and with higher quality, achieving the effects of key attention, full-staff supervision, and efficient handling.

[0067] Embodiment 4

[0068] This embodiment is basically the same as the above embodiments, except that it further includes: setting user permissions and problem permissions; users can view problems corresponding to the problem permissions according to their user permissions.

[0069] Specifically, for setting user permissions and problem permissions, in this embodiment, the user permissions and problem permissions are set according to the job ranks of enterprise employees. For example, the job ranks range from 1 to 5, and the larger the value, the higher the job rank. The corresponding user permissions and problem permissions also range from 1 to 5. A user with a user permission of 4 can view problems with problem permissions of 1 - 4. That is, their display module will only display problems with problem permissions of 1 - 4 and their evaluation information, thus facilitating hierarchical management of problems and confidentiality of relevant content.

[0070] The above are only embodiments of the present invention. Common knowledge such as specific structures and characteristics known in the art is not described in detail here. 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 conventional experimental means before this date. Those of ordinary skill in the art can, under the inspiration given in this application, complete and implement this solution in combination with their own abilities. Some typical well-known structures or well-known methods should not be 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, which 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 based on the content of its claims, and the specific implementation manners described in the specification can be used to interpret the content of the claims.

Claims

1. An AI learning system based on problem supervision, characterized in that: include: The training module is used to obtain the whole process data of the problem as a training data set to train the AI ​​model, including: Obtain the whole process data of the problem, take all the whole process data except the evaluation data as input, and the evaluation data as output, and use the positive evaluation and negative evaluation in the evaluation data as data annotation; The training data set formed by input and output is used to train the AI ​​model so that the AI ​​model can analyze and evaluate the problem; The full process data of the problem includes: all data generated from the initiation to the resolution of the problem, as well as subsequent evaluation data. The evaluation data includes: one or more of the evaluation information of the problem initiator, the evaluation information of the supervisory leader, and the evaluation information of the public problem.

2. The problem-based AI learning system according to claim 1, characterized in that: The training module is used to obtain the whole process data of the problem as a training data set to train the AI ​​model, and also includes: Extract part of the input and output to form a test set, use the test set to evaluate the performance of the trained model, and adjust the AI ​​model parameters or structure based on the evaluation results.

3. The problem-based AI learning system according to claim 1, characterized in that: Also includes: Information acquisition module, display module, user module, evaluation module; The information acquisition module is used to obtain public selection information of issues to be resolved, selection information of issues to be supervised, and evaluation information of public issues; A display module is used to publicly select information based on the problem to be solved, and obtain the problem to be solved for public disclosure; The user module is used to select information based on the issues that the user is concerned about and supervises, obtain the corresponding issues, and add them to the user's concerned and supervised list; The evaluation module is used to receive evaluation information of public questions and display it in association with the public questions.

4. The problem-based AI learning system according to claim 1, characterized in that: Also includes: permission management module; The permission management module is used to set user permissions and question permissions; users can view questions with corresponding question permissions based on user permissions.

5. An AI learning method based on problem supervision, characterized in that: include: Obtain the entire process data of the problem as a training data set to train the AI ​​model, including: Obtain the whole process data of the problem, take all the whole process data except the evaluation data as input, and the evaluation data as output, and use the positive evaluation and negative evaluation in the evaluation data as data annotation; The training data set formed by input and output is used to train the AI ​​model so that the AI ​​model can analyze and evaluate the problem; The full process data of the problem includes: all data generated from the initiation to the resolution of the problem, as well as subsequent evaluation data. The evaluation data includes: one or more of the evaluation information of the problem initiator, the evaluation information of the supervisory leader, and the evaluation information of the public problem.

6. The problem-based AI learning method according to claim 5, characterized in that: The whole process data of the problem is obtained as a training data set to train the AI ​​model, and further includes: Extract part of the input and output to form a test set, use the test set to evaluate the performance of the trained model, and adjust the AI ​​model parameters or structure based on the evaluation results.

7. The problem-based AI learning method according to claim 5, characterized in that: Also includes: Obtain public selection information on pending issues, selection information on issues of concern and supervision, and evaluation information on public issues; Select information based on the issues to be resolved and obtain the issues to be resolved for disclosure; Select information based on the issues you are concerned about and supervising, obtain the corresponding issues, and add them to the user's concerned and supervising list; Receive evaluation information of public questions and display it in association with the public questions.

8. The problem-based AI learning method according to claim 5, characterized in that: Also includes: Set user permissions and question permissions; Users can view questions with corresponding permissions based on their user permissions.

Images