Student autorotation risk assessment method based on large model

Abstract

The invention discloses a student autorotation risk assessment method based on a large model. The method specifically comprises the following steps: S1, establishing an assessment model; s2, big data training; s3, limiting parameter setting; s4, evaluating standard data; s5, acquiring text data; s6, importing the text content; the method has the advantages that the data is selected from the big data to serve as a model training set, so that the accuracy of the model is improved, meanwhile, the text scanning frequency, synonym setting and limited parameters of special vocabulary fuzzy analogy are set, the learning ability of the model is further improved while the working efficiency of the model is improved, and the method is suitable for popularization and application. And finally, a multi-dimensional result of a model evaluation report is set, so that the evaluation result can be better displayed, the evaluation result is compared with a manual evaluation result, and when the analogy does not pass, the manual evaluation result is imported as a learning index for learning again, so that the learning ability of the model is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of model evaluation, and in particular to a method for student autobiography risk evaluation based on a large model. Background Art

[0002] Students' primary living environment is the campus, and their primary task is learning. Furthermore, they live in an era of rapid social development, change, and competition. They experience numerous psychological conflicts and setbacks related to their studies, campus life, and social change. Understanding and grasping students' unique psychological backgrounds can help scientifically examine the hierarchical nature of their psychological problems. Based on the current state of Chinese schools and the psychological realities of students, students' psychological problems can generally be categorized into three levels: psychological deviations, psychological disorders, and mental illnesses.

[0003] A person's mental state will unconsciously flow into his or her daily words and deeds. Students are no exception. In students' daily records, insights, articles and other personal written documents, their mental state will be revealed involuntarily. In order to better control the mental health level of students, a method of student autobiography risk assessment based on a large model is proposed. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for assessing the risk of student autobiographies based on a large model. Taking student autobiographies as the starting point, the method analyzes these text languages ​​through the large model to obtain the student's mental health level, so as to solve the problems raised in the above background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A method for assessing student autobiography risk based on a large model, specifically comprising the following steps:

[0007] S1. Establish an assessment model: Establish a psychological safety assessment model based on a large model;

[0008] S2. Big Data Training: Retrieve and obtain text and corresponding mental health-related data from big data, perform feature engineering, and then import them into the model for training;

[0009] S3, limited parameter setting: setting safety measures in the established model;

[0010] S4. Standard data evaluation: retrieve standard data corresponding to different types of psychological problems and import them into the trained model for evaluation;

[0011] S5. Obtain text data: Obtain the autobiographical text data of the student to be evaluated;

[0012] S6. Text content import: Conduct a preliminary analysis of the text content, screen out obvious sensitive words, and import the complete autobiographical text data into the determined model;

[0013] S7. Export evaluation report: Export the evaluation report generated by the model and provide feedback and optimization.

[0014] As a preferred solution of the present invention, the step of establishing the evaluation model in S1 specifically includes the following steps:

[0015] S1.1. Select the standard psychological assessment report template as the evaluation target and the BLEU evaluation indicator;

[0016] S1.2. Build a system architecture for the psychological safety assessment model through a modular structure;

[0017] S1.3. Compile the model’s interface functions and evaluation functions;

[0018] S1.4. Perform model evaluation and corresponding optimization.

[0019] As a preferred solution of the present invention: during the training process of S2, the correspondence between the text data and the psychological assessment report needs to be ensured, and the collected data needs to be cleaned and sorted, and word segmentation and tokenization processing needs to be performed at the same time.

[0020] As a preferred solution of the present invention: the limited parameters in S3 include the number of text scans, synonym settings, and fuzzy analogies of special words.

[0021] As a preferred solution of the present invention, the standard data evaluation of S4 specifically includes the following steps:

[0022] S4.1. Select 3-5 text data and their corresponding standard evaluation results as test data;

[0023] S4.2, import the text data of the selected test data into the model for evaluation;

[0024] S4.3. Record the response time and resource consumption of the model evaluation and export the evaluation report;

[0025] S4.4. Compare the evaluation report derived from the model with the standard evaluation results and present the comparison results;

[0026] S4.5. Optimize the model based on the analogy results.

[0027] As a preferred solution of the present invention: the student autobiographical text data in S5 needs to ensure that the content is complete and not tampered with. The acquisition of the autobiographical text data can be divided into student self-submission, teacher-assisted submission and student text system export.

[0028] As a preferred solution of the present invention: the analysis of the text content in S6 includes inappropriate speech screening, sensitive information screening, plagiarism information screening and implicit information extraction, and the analysis results need to be exported into a corresponding report, and manual evaluation and analysis are performed at the same time to obtain a manual evaluation and analysis report.

[0029] As a preferred solution of the present invention: the model analysis report exported by S7 is set to report the main results, branch results, result traceability and improvement measures, the main results include normal mental health status, mild psychological distress, toxic psychological disorder and severe psychological crisis, and the branch results include inappropriate speech count, sensitive word count and plagiarism information count.

[0030] As a preferred solution of the present invention: the model analysis report needs to be compared with the manual evaluation analysis report, and analogies are made in terms of main results, branch results, result traceability and improvement measures. At the same time, the analogy results are given. For model analysis reports with a comparison analogy similarity of less than 90%, the manual analysis report needs to be imported into the model for re-learning.

[0031] Compared with the existing technology, the beneficial effects of the present invention are: selecting data from big data as a model training set, thereby improving the accuracy of the model, and at the same time setting the limit parameters of the number of text scanning times, synonym settings, and fuzzy analogies of special vocabulary, while improving the working efficiency of the model, further improving the learning ability of the model, and improving the evaluation accuracy of the model, and finally setting the multi-dimensional results of the model evaluation report, which can better display the evaluation results and compare them with the manual evaluation results. When the analogy fails, the manual evaluation results are imported as learning indicators for re-learning to improve the learning ability of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is a system block diagram of the present invention. DETAILED DESCRIPTION

[0033] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0034] See also Figure 1 , the present invention provides a technical solution:

[0035] A method for assessing student autobiography risk based on a large model, specifically comprising the following steps:

[0036] S1. Establish an assessment model: Establish a psychological safety assessment model based on a large model;

[0037] S2. Big Data Training: Retrieve and obtain text and corresponding mental health-related data from big data, perform feature engineering, and then import them into the model for training;

[0038] S3, limited parameter setting: setting safety measures in the established model;

[0039] S4. Standard data evaluation: retrieve standard data corresponding to different types of psychological problems and import them into the trained model for evaluation;

[0040] S5. Obtain text data: Obtain the autobiographical text data of the student to be evaluated;

[0041] S6. Text content import: Conduct a preliminary analysis of the text content, screen out obvious sensitive words, and import the complete autobiographical text data into the determined model;

[0042] S7. Export evaluation report: Export the evaluation report generated by the model and provide feedback and optimization.

[0043] Furthermore, the establishment of the evaluation model in S1 specifically includes the following steps:

[0044] S1.1. Select the standard psychological assessment report template as the evaluation target and the BLEU evaluation indicator;

[0045] S1.2. Build a system architecture for the psychological safety assessment model through a modular structure;

[0046] S1.3. Compile the model’s interface functions and evaluation functions;

[0047] S1.4. Perform model evaluation and corresponding optimization.

[0048] Furthermore, during the training process of S2, the correspondence between the text data and the psychological assessment report needs to be ensured, and the collected data needs to be cleaned and organized, and word segmentation and tokenization processing needs to be performed at the same time.

[0049] Furthermore, the limited parameters in S3 include the number of text scans, synonym settings, and fuzzy analogies of special words.

[0050] Furthermore, the standard data evaluation of S4 specifically includes the following steps:

[0051] S4.1. Select 3-5 text data and their corresponding standard evaluation results as test data;

[0052] S4.2, import the text data of the selected test data into the model for evaluation;

[0053] S4.3. Record the response time and resource consumption of the model evaluation and export the evaluation report;

[0054] S4.4. Compare the evaluation report derived from the model with the standard evaluation results and present the comparison results;

[0055] S4.5. Optimize the model based on the analogy results.

[0056] Furthermore, the student autobiographical text data in S5 needs to ensure that the content is complete and has not been tampered with. The acquisition of the autobiographical text data can be divided into student self-submission, teacher-assisted submission and student text system export.

[0057] Furthermore, the analysis of the text content in S6 includes inappropriate speech screening, sensitive information screening, plagiarism information screening and implicit information extraction. The analysis results need to be exported into a corresponding report, and manual evaluation and analysis are performed at the same time to obtain a manual evaluation and analysis report.

[0058] Furthermore, the model analysis report exported by S7 is set to report the main results, branch results, result traceability and improvement measures. The main results include normal mental health status, mild psychological distress, toxic psychological disorder and severe psychological crisis. The branch results include inappropriate speech counts, sensitive word counts and plagiarized information counts.

[0059] Furthermore, the model analysis report needs to be compared with the manual evaluation analysis report, and analogies should be made in terms of main results, branch results, result traceability and improvement measures. At the same time, the analogy results should be given. For model analysis reports with a comparison and analogy similarity of less than 90%, the manual analysis report needs to be imported into the model for re-learning.

[0060] The above technical solution mentions a complete student autobiography risk assessment method. The specific steps of this method are as follows:

[0061] 1. Obtain student autobiographical content: First, you need to obtain the student's autobiographical content. This can be done through online platforms, email, or other electronic means. Ensure that the content obtained is complete and has not been tampered with.

[0062] Content Analysis and Risk Identification: Leveraging large models, content analysis can identify potential risks. This includes checking for inappropriate language, sensitive information leaks, plagiarism, and other issues. Large models can analyze text sentiment, keyword frequency, and other factors to help identify potential risks.

[0063] Set safety measures: Build safety measures into the model, such as limiting the number and duration of daily conversations for minors, protecting personal information, and providing channels for complaints and reporting. These measures ensure that the model does not leak sensitive information or generate harmful content during operation.

[0064] Performance and stability evaluation: Test the model's response time, resource consumption, and stability to ensure that the model can complete tasks within a reasonable time and utilizes resources appropriately. This includes long-running tests and large-scale data input tests to verify the model's stability and performance.

[0065] Feedback and optimization: Feedback and optimization are provided based on the evaluation results. If the model performs poorly in some aspects, the performance can be improved by adjusting model parameters or adding more training data.

[0066] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for assessing student autobiography risk based on a large model, characterized in that: The specific steps include: S1. Establish an assessment model: Establish a psychological safety assessment model based on a large model; S2. Big data training: retrieve and obtain texts and corresponding mental health-related data in big data, perform feature engineering, and then import them into the model for training; S3, limited parameter setting: setting safety measures in the established model; S4. Standard data evaluation: retrieve standard data corresponding to different types of psychological problems and import them into the trained model for evaluation; S5. Obtaining text data: obtaining the autobiographical text data of the student to be evaluated; S6, text content import: conduct preliminary analysis of the text content, screen out obvious sensitive words, and import the complete autobiographical text data into the determined model; S7. Evaluation report export: Export the evaluation report generated by the model and provide feedback and optimization.

2. The method for assessing student autobiography risk based on a large model according to claim 1, characterized in that: The establishment of the evaluation model in S1 specifically includes the following steps: S1.

1. Select the standard psychological assessment report template as the evaluation target and select the BLEU evaluation indicator; S1.

2. Build the system architecture of psychological safety assessment model through modular structure; S1.3, compile the interface function and evaluation function of the model; S1.

4. Perform model evaluation and make corresponding optimizations.

3. The method for student autobiography risk assessment based on a large model according to claim 1 is characterized by: During the training process of S2, the correspondence between the text data and the psychological assessment report needs to be ensured. At the same time, the collected data needs to be cleaned and organized, and word segmentation and tokenization processing needs to be performed at the same time.

4. The method for assessing student autobiography risk based on a large model according to claim 1, characterized in that: The limited parameters in S3 include text scanning times, synonym settings, and special vocabulary fuzzy analogies.

5. The method for student autobiography risk assessment based on a large model according to claim 1 is characterized by: The standard data evaluation of S4 specifically includes the following steps: S4.

1. Select 3-5 items of text data and their corresponding standard evaluation results as test data; S4.2, importing the text data of the selected test data into the model for evaluation; S4.

3. Record the response time and resource consumption of the model evaluation and export the evaluation report; S4.

4. Compare the evaluation report derived from the model with the standard evaluation results and give the comparison results; S4.

5. Optimize the model based on the analogy results.

6. The method for student autobiography risk assessment based on a large model according to claim 1 is characterized by: The student autobiography text data in S5 needs to ensure that the content is complete and not tampered with. The acquisition of the autobiography text data can be divided into student self-submission, teacher-assisted submission and student text system export.

7. The method for student autobiography risk assessment based on a large model according to claim 1 is characterized by: The analysis of the text content in S6 includes inappropriate speech screening, sensitive information screening, plagiarized information screening and implicit information extraction. The analysis results need to be exported to a corresponding report, and manual evaluation and analysis are performed at the same time to obtain a manual evaluation and analysis report.

8. The method for student autobiography risk assessment based on a large model according to claim 7 is characterized by: The model analysis report exported by S7 is set to report the main results, branch results, result traceability and improvement measures. The main results include normal mental health status, mild psychological distress, toxic psychological disorders and severe psychological crisis. The branch results include inappropriate speech counts, sensitive word counts and plagiarized information counts.

9. The method for student autobiography risk assessment based on a large model according to claim 8, characterized in that: The model analysis report needs to be compared with the manual evaluation analysis report, and analogies should be made in terms of main results, branch results, result traceability and improvement measures. At the same time, the analogy results should be given. For model analysis reports with a comparison analogy similarity of less than 90%, the manual analysis report needs to be imported into the model for re-learning.

Images