Higher education evaluation system and method based on big data technology

Abstract

The invention discloses a higher education evaluation system and method based on a big data technology, and belongs to the technical field of big data, and the system comprises the steps: S1, uploading the attendance card punching information of college teachers and students, and uploading the daily examination score information and test paper examination questions of the college students; s2, score classification and test paper difficulty assessment; s3, performing comprehensive evaluation on college teachers according to attendance card punching information of the college teachers and students, uploading daily examination score information of the college students and test paper examination questions; according to the method, examination paper difficulty evaluation is set, examination question difficulty is analyzed according to imported examination questions of the examination paper, the P value (namely the proportion of the number of persons answering the questions) of each question is calculated according to the answering condition of examinees, the P values of the multiple questions of the examination paper are added to calculate the average P value, and if the average P value is remarkably lower than the expectation, it is indicated that the questions are difficult, and if not, the questions are not difficult. The system judges that the scores of the single subjects are invalid, the scores are not included in teacher evaluation reference, and the problem that the accuracy of teacher evaluation is affected due to the fact that examination questions are difficult and not teacher teaching reasons is avoided.

Description

Technical Field

[0001] The present invention belongs to the field of big data technology, and specifically relates to a higher education evaluation system and method based on big data technology. Background Art

[0002] Data sets with large capacity, multiple types, fast access speed and high application value were first applied in the IT industry. They are now rapidly developing into a new generation of information technology and service formats that collect, store and analyze huge amounts of data with scattered sources and diverse formats, discover new knowledge, create new value and enhance new capabilities. Big data must adopt a distributed architecture to conduct distributed data mining on massive data. Therefore, it must rely on distributed processing, distributed databases and cloud storage, and virtualization technology of cloud computing. With the continuous development of higher vocational education, education quality evaluation has become an important part of education management. Teacher evaluation is the process of evaluating and assessing teachers' teaching performance, professional quality, and teaching achievements. Its purpose is usually to improve teaching quality, promote teachers' professional development, motivate teachers' work performance, and provide decision-making basis for school management.

[0003] The existing technology has the following problems: 1. The existing higher education evaluation methods based on big data technology are unable to classify the imported scores into corresponding teachers and evaluate the difficulty of the test questions. When the test questions are too difficult, the scores cannot be invalidated. When the difficulty of the test questions is greater than the difficulty of daily teaching, there will be errors in the test scores used as teacher evaluations. 2. The existing higher education evaluation methods based on big data technology cannot encrypt the system's evaluation of teachers. Teachers in colleges and universities can view other teachers' evaluation information, which has poor confidentiality; Summary of the invention

[0004] In order to solve the problems raised in the above background technology, the present invention provides a higher education evaluation system and method based on big data technology, which has the characteristics of invalidating the results when the test questions are too difficult, thereby improving the accuracy of teacher evaluation.

[0005] To achieve the above object, the present invention provides the following technical solution: a higher education evaluation method based on big data technology, comprising the following steps: S1: Upload the attendance information of teachers and students of the school, upload the daily assessment score information and test questions of students of the school; S2: classification of scores and assessment of test difficulty; S3: Attendance check-in information of teachers and students in colleges and universities, uploading daily assessment score information and test papers of college students to conduct comprehensive evaluation of college teachers; S4: Evaluation encryption; S5: Output comments.

[0006] Furthermore, the grade classification includes the following steps: Based on course code: If the grade file contains a course code, the system can directly query the corresponding course in the database based on the course code and obtain the information of the main teacher of the course; Based on teacher ID: Some systems allow teacher IDs to be specified directly in the grade file, so that even if there are multiple teachers teaching the course, the grade can be accurately assigned to each teacher; Multi-condition matching: For complex scenarios, such as when a course is taught by multiple teachers in different classes, the system can achieve more accurate matching by combining multiple conditions such as course code and class number.

[0007] Furthermore, the difficulty assessment of the test paper includes the following steps: Test paper and question import: import all the contents of the test paper and question into the system; Difficulty analysis of test questions: evaluate the difficulty of test questions; Comparison: Analyze the difficulty of the test questions based on the imported test questions, calculate the P value of each question (that is, the proportion of people who answered the question correctly) according to the candidates' answers, and add up the P values ​​of multiple questions in the test paper to calculate the average P value. If the average P value is significantly lower than expected, it means that the question is too difficult. The system determines that the score of the single subject is invalid, and the score is not included in the teacher's evaluation reference to avoid the accuracy of the teacher's evaluation being affected by the difficulty of the test questions that are not due to the teacher's teaching. On the contrary, when the P value is not lower than the preset value, the score is included in the teacher's evaluation reference.

[0008] Furthermore, the evaluation encryption includes the following steps: Data extraction: extract teacher-related information from the database based on the corresponding teachers evaluated; Data encryption: Encrypt the evaluation data based on teacher-related information; Generate key: Encrypt the evaluation information and generate questions. After passing the questions, you can view the evaluation information of the relevant teachers.

[0009] Furthermore, the course code is based on writing SQL queries or using query methods in the ORM framework.

[0010] Furthermore, the difficulty of the test questions is evaluated by using an item response theory (IRT) algorithm.

[0011] Furthermore, the difficulty comparison of the test questions is performed using a decision tree and a support vector machine (SVM).

[0012] Furthermore, the data extraction uses natural language processing (NLP) to extract teacher-related information from a database based on the corresponding teachers evaluated.

[0013] Furthermore, after the key is generated, a digital signature or date of birth can be generated based on the teacher's information to unlock and view the evaluation information.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention sets the grade classification, and can match the corresponding teacher according to the imported subject grades, so as to improve the accuracy of teacher evaluation. By setting the test paper difficulty assessment, the test question difficulty is analyzed according to the imported test paper questions, and the P value of each question (that is, the proportion of the number of people who answer the question correctly) is calculated according to the test takers' answers. The P values ​​of multiple questions in the test paper are added to calculate the average P value. If the average P value is significantly lower than expected, it indicates that the question is too difficult. The system determines that the single subject score is invalid, and the score is not included in the teacher evaluation reference, so as to avoid the accuracy of teacher evaluation being affected by the difficulty of the test question which is not due to the teacher's teaching. On the contrary, when the P value is not lower than the preset value, the score is included in the teacher evaluation reference.

[0015] 2. The present invention sets evaluation encryption. After the system completes the teacher evaluation, the system can generate a key based on the teacher information corresponding to the evaluation information. When you want to view the evaluation information of a certain teacher, you need to pass the key question to view the evaluation information, thereby improving the confidentiality of the system's teacher evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flowchart of the present invention; Figure 2 It is a flowchart of the score classification process of the present invention; Figure 3 It is a flowchart of the test paper difficulty assessment process of the present invention; Figure 4 Evaluation of encryption flow chart for the present invention; DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0018] See also Figure 1-4 , the present invention provides the following technical solution: comprising the following steps: S1: Upload the attendance information of teachers and students of the school, upload the daily assessment score information and test questions of students of the school; S2: classification of scores and assessment of test difficulty; S3: Attendance check-in information of teachers and students in colleges and universities, uploading daily assessment score information and test papers of college students to conduct comprehensive evaluation of college teachers; S4: Evaluation encryption; S5: Output comments.

[0019] Further in the present invention, the grade classification comprises the following steps: Based on course code: If the grade file contains a course code, the system can directly query the corresponding course in the database based on the course code and obtain the information of the main teacher of the course; Based on teacher ID: Some systems allow teacher IDs to be specified directly in the grade file, so that even if there are multiple teachers teaching the course, the grade can be accurately assigned to each teacher; Multi-condition matching: For complex scenarios, such as when a course is taught by multiple teachers in different classes, the system can achieve more accurate matching by combining multiple conditions such as course code and class number.

[0020] By adopting the above technical solution, the corresponding teachers can be matched according to the imported subject grades. At the same time, multi-condition matching can be carried out for the situation where the same course is taught by multiple teachers in different classes, thereby avoiding errors in teacher evaluation caused by confusion in the imported grades and improving the accuracy of teacher evaluation.

[0021] Further in the present invention, the test paper difficulty assessment comprises the following steps: Test paper and question import: import all the contents of the test paper and question into the system; Difficulty analysis of test questions: evaluate the difficulty of test questions; Comparison: Analyze the difficulty of the test questions based on the imported test questions, calculate the P value of each question (that is, the proportion of people who answered the question correctly) according to the candidates' answers, and add up the P values ​​of multiple questions in the test paper to calculate the average P value. If the average P value is significantly lower than expected, it means that the question is too difficult. The system determines that the score of the single subject is invalid, and the score is not included in the teacher's evaluation reference to avoid the accuracy of the teacher's evaluation being affected by the difficulty of the test questions that are not due to the teacher's teaching. On the contrary, when the P value is not lower than the preset value, the score is included in the teacher's evaluation reference.

[0022] By adopting the above technical solution and setting the test paper difficulty assessment, the test question difficulty is analyzed according to the imported test paper questions. When the test question difficulty is greater than the preset difficulty, the score is invalid, so as to avoid the accuracy of teacher evaluation being affected by the high difficulty of the test question which is not due to the teacher's teaching.

[0023] Further in the present invention, the evaluation encryption comprises the following steps: Data extraction: extract teacher-related information from the database based on the corresponding teachers evaluated; Data encryption: Encrypt the evaluation data based on teacher-related information; Generate key: Encrypt the evaluation information and generate questions. After passing the questions, you can view the evaluation information of the relevant teachers.

[0024] By adopting the above technical solution and setting up evaluation encryption, after the system completes the teacher evaluation, the system can generate a key based on the teacher information corresponding to the evaluation information. When you want to view the evaluation information of a certain teacher, you need to pass the key question to view the evaluation information, thereby improving the confidentiality of the system's teacher evaluation.

[0025] Furthermore, in the present invention, SQL queries are written based on the course code or query methods in the ORM framework are used.

[0026] By adopting the above technical solutions, SQL provides great flexibility. For complex queries, such as multi-table connections, aggregate functions, window functions, etc., SQL can often provide more efficient solutions. Since SQL statements are directly executed by the database engine, they usually have better performance when processing large-scale data. Especially in optimized queries, SQL can make full use of database indexes and other optimization mechanisms to reduce I / O overhead and CPU consumption. SQL queries can be run directly in database client tools, which is convenient for testing and debugging. In addition, many database systems also provide detailed execution plans to help developers analyze query performance and find potential problems. The ORM framework automatically handles the mapping relationship between objects and database tables, reducing the workload of writing and maintaining SQL statements. Using the ORM framework can significantly reduce the time for writing and maintaining database access code. Traditional data access methods usually require developers to manually write a large number of SQL statements to implement operations such as adding, deleting, modifying, and checking. The ORM framework allows developers to directly operate entity classes in an object-oriented manner by providing high-level abstractions, thereby reducing repetitive code writing work. For example, in Java, developers can complete database operations through simple API calls without having to worry about the underlying SQL syntax. The ORM framework hides the specific database access details, allowing developers to focus on the implementation of business logic. It provides a unified set of interfaces, and developers can use the same API to operate regardless of which relational database (such as MySQL, PostgreSQL, Oracle, etc.) is used in the background. This abstraction not only simplifies the development process, but also enhances the portability of the code, because when changing the database, only the connection information in the configuration file needs to be adjusted without modifying the application code. The ORM framework makes unit testing easier because it allows developers to simulate data access behavior without a real database. By using an in-memory database or mock object, developers can quickly run test cases locally to ensure the correctness of the business logic. In addition, the ORM framework usually provides transaction management functions, which can automatically roll back changes during the test process to ensure the consistency of the test environment. Many ORM frameworks have built-in caching mechanisms that can effectively reduce the number of database accesses and improve the system's response speed. For example, the first-level cache (session-level cache) will save loaded objects in the same transaction to avoid repeated queries; the second-level cache (application-level or cluster-level cache) can share data in a wider range to further improve performance. In addition, the ORM framework can also be combined with other optimization technologies, such as lazy loading and eager loading, to select the most appropriate data loading strategy according to actual needs.

[0027] In the present invention, the difficulty of the test questions is further evaluated by using an item response theory (IRT) algorithm.

[0028] By adopting the above technical solutions, in the IRT framework, the estimation of the parameters of the questions is independent of the sample of the subjects, which means that the results between different samples can be easily converted to the same scale through linear transformation. This is in sharp contrast to CTT, whose results are highly dependent on specific sample characteristics. Therefore, IRT can provide more stable and consistent ability estimates, which are not affected by specific sample characteristics. The IRT model is not only applicable to single-form tests, but also can handle sparse matrix problems introduced by multiple forms, linear dynamic tests or adaptive tests. This flexibility makes IRT very suitable for the development of computerized adaptive tests (CATs), in which subsequent questions are dynamically selected based on the performance of the subjects, thereby improving measurement accuracy and shortening test time. When the exam has multiple forms or a new form is used every year, IRT provides a stronger equivalence method than CTT, ensuring that the scores between different forms are more comparable and valid. This is particularly important for tracking student progress over the long term or making comparisons across years. Classic tests are usually designed for average students and perform poorly for high- or low-scoring students. IRT, on the other hand, can provide accurate ability estimates across a wide range of latent trait levels. Even very difficult or very easy questions can be reasonably described statistically. In addition, the information function of IRT can clearly show the amount of information a test contains at different ability levels, helping to identify which questions are best suited for students within a specific ability range.

[0029] Furthermore, in the present invention, the difficulty of the test questions is compared using a decision tree and a support vector machine (SVM).

[0030] By adopting the above technical solutions, an important feature of the decision tree is that its results are easy to understand and explain. The tree structure intuitively shows the decision process. Each node represents a test on an attribute, each branch represents a possible attribute value, and each leaf node represents the final classification result or numerical prediction. This visual expression allows non-professionals to easily understand the working principle of the model, which is particularly important for scenarios where model results need to be explained to business personnel or customers. Decision trees can handle both discrete and continuous data, which makes it more flexible when facing complex data sets. For discrete data, decision trees can directly use attribute values ​​for division; for continuous data, decision trees can achieve effective segmentation by setting thresholds. In addition, decision trees can also handle missing values, by ignoring samples with missing values ​​or using specific strategies to fill in missing values, thereby ensuring the robustness of the model. Compared with other complex machine learning models, decision trees are faster to train. This is because the construction process of the decision tree is mainly based on a greedy algorithm, that is, the optimal splitting attribute is selected at each step to maximize information gain or minimize Gini impurity. Although this method cannot guarantee the global optimal solution, it can quickly generate a tree with good performance in most cases. In addition, decision trees do not require complex preprocessing of input data, such as standardization or normalization, which further speeds up the training process. Decision trees have a certain tolerance for noisy data. When there are outliers or noise in the data set, decision trees can avoid overfitting by setting appropriate pruning strategies to ensure that the model has good generalization ability. For example, the C4.5 algorithm introduces the information gain rate to overcome the problem that information gain tends to select attributes with many values, and prunes during the tree construction process to reduce the risk of overfitting. SVM is based on statistical learning theory, especially VC dimension theory and structural risk minimization principle. These theories ensure that SVM can find the optimal solution in the case of limited samples and has good generalization ability. Unlike traditional probability measurement-based methods, SVM avoids the process from induction to deduction and realizes efficient "transductive reasoning", that is, directly inferring the category of unknown samples from training samples12. This means that SVM can effectively handle small sample data sets while maintaining high classification accuracy. SVM constructs the optimal decision boundary by maximizing the classification interval (Margin), thereby improving the model's prediction accuracy for new data. Specifically, SVM looks for a hyperplane that can separate data points of different categories as much as possible, so that the distance between the data points closest to the hyperplane is maximized. This design not only enhances the robustness of the model, but also makes it more tolerant to noise and outliers.In addition, the generalization ability of SVM is also reflected in its ability to process data in high-dimensional space. Even if the number of features is much larger than the number of samples, it will not lead to overfitting problems. When faced with nonlinearly separable data, SVM can introduce a kernel function to map the original input space to a higher-dimensional space, where the data may become linearly separable. Commonly used kernel functions include linear kernels, polynomial kernels, radial basis functions (RBF), etc. This mechanism allows SVM to handle complex pattern recognition tasks without increasing computational complexity. For example, in the field of image recognition, SVM combined with appropriate kernel functions can achieve effective classification of various objects. In SVM, only those sample points located near the decision boundary, that is, support vectors, will affect the final classifier construction. This shows that SVM can focus on the most valuable information and ignore redundant data far away from the decision boundary. Therefore, SVM can not only simplify the model structure, but also improve computational efficiency, especially on large-scale data sets11. In addition, since the number of support vectors is usually small, this also means that SVM has high stability to changes such as adding or removing non-critical samples, and the computational complexity of SVM depends mainly on the number of support vectors rather than the dimension of the feature space. This property makes SVM very suitable for processing high-dimensional data sets, because even if the number of features is very large, as long as the number of support vectors remains relatively small, it will not significantly increase the computational burden. This is particularly important for fields such as text classification and gene expression analysis, where the feature dimensions are often very high.

[0031] Furthermore, in the present invention, data extraction uses natural language processing (NLP) to extract teacher-related information from a database based on the corresponding teachers evaluated.

[0032] By adopting the above technical solutions, NLP allows people to communicate with computers and digital devices in a more natural way, breaking the limitations of the traditional human-computer interaction mode based on command lines or graphical interfaces. Through technologies such as speech recognition, semantic understanding, and dialogue management, users can ask questions or issue instructions in natural language, and the system can accurately interpret the intention and respond. For example, in products such as smart speakers and virtual assistants, NLP makes human-computer interaction smoother and more intuitive. NLP has a powerful text generation capability and can automatically generate high-quality articles, summaries, reports and other content based on a given topic or context. This automated content creation not only improves production efficiency, but also ensures the consistency and accuracy of information. In addition, NLP can also be used in personalized recommendation systems to generate customized news, advertisements or other forms of information based on user preferences. NLP can automatically extract valuable information such as time, place, people, events, etc. from a large amount of unstructured text data and convert it into a structured format. This technology is particularly important for enterprises because it can help them quickly obtain key business information and support decision making. For example, in the financial industry, NLP can be used to analyze market dynamics and monitor risk factors; in the healthcare field, it can be used for medical record management and disease diagnosis assistance. The intelligent question-answering system based on NLP can understand the natural language questions raised by users, retrieve relevant information from the knowledge base or the Internet, and give accurate answers. Such systems are widely used in customer service robots, online education platforms and other fields, effectively reducing the workload of manual customer service and improving user experience. For example, the automatic reply function on the e-commerce platform can use NLP technology to handle a large number of repetitive questions, so that customers can get timely help. NLP is one of the important tools for building knowledge graphs. It can extract entities and their relationships from text and organize them into a structured network. Such knowledge representation is not only convenient for query and reasoning, but also provides a solid foundation for other AI applications. For example, in intelligent search, knowledge graphs can enable search engines to better understand users' query intentions and return more relevant results; in recommendation systems, it can enhance the effect of personalized recommendations. Although this is not an inherent advantage of NLP, in the current environment that emphasizes data security and personal privacy protection, NLP technology can also play a role in this field. For example, through desensitization and anonymization, NLP can ensure that sensitive information is not leaked without affecting the analysis effect. At the same time, NLP can also be used for security-related tasks such as detecting malicious content and preventing network attacks. Chatbots and voice assistants supported by NLP technology have become a key means to improve customer service experience. They can be online 24 / 7, respond to customer inquiries instantly, solve common problems, and even complete certain transaction processes. This not only improves service efficiency, but also reduces operating costs.In addition, NLP can also be used to analyze customer feedback, helping companies to continuously improve product quality and service levels. NLP is also good at discovering potential patterns and trends from massive text data, including cluster analysis, classification prediction, topic modeling and other tasks. These technologies help to reveal the knowledge hidden behind the data and provide support for scientific research, business intelligence and other activities. For example, in public opinion monitoring, NLP can help government agencies and social organizations to grasp the focus of public attention and emotional changes in a timely manner by collecting and processing online public opinion, so as to take corresponding measures.

[0033] Furthermore, in the present invention, after the key is generated, a digital signature or date of birth and other information can be generated based on the teacher's information to unlock and view the evaluation information.

[0034] By adopting the above technical solution, a digital signature or questions such as date of birth are generated according to the teacher's information to unlock and view the evaluation information, making the generated questions more private and thus more secure.

[0035] The working principle and use process of the present invention are as follows: the higher education evaluation method based on big data technology, which uploads the attendance punch-in information of teachers and students of the college, uploads the daily assessment score information and test paper questions of the students of the college, and then classifies the scores and evaluates the difficulty of the test paper. According to the attendance punch-in information of teachers and students of the college, the daily assessment score information and test paper questions of the students of the college are uploaded to comprehensively evaluate the teachers of the college, and the comments are output after the evaluation information is encrypted. By setting the score classification, the corresponding teachers can be matched according to the imported subject scores, thereby improving the accuracy of the evaluation of teachers. By setting the test paper difficulty evaluation, the difficulty of the test paper is analyzed according to the imported test paper questions, and the answer of each test paper is calculated according to the test taker's answer. The P value of a question (i.e. the proportion of people who answered the question correctly) is calculated, and the P values ​​of multiple questions in the test paper are added to calculate the average P value. If the average P value is significantly lower than expected, it means that the question is too difficult. The system determines that the score of the single subject is invalid, and the score is not included in the teacher evaluation reference to avoid the accuracy of the teacher evaluation being affected by the difficulty of the test questions not being the result of the teacher's teaching. On the contrary, when the P value is not lower than the preset value, the score is included in the teacher evaluation reference. By setting evaluation encryption, after the system completes the teacher evaluation, the system can generate a key based on the teacher information corresponding to the evaluation information. When you want to view the evaluation information of a certain teacher, you need to pass the key question to view the evaluation information, thereby improving the confidentiality of the system's teacher evaluation.

[0036] Although 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 the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A higher education evaluation method based on big data technology, characterized by: The steps include: S1: Upload the attendance information of teachers and students of the school, upload the daily assessment score information and test questions of students of the school; S2: classification of scores and assessment of test difficulty; S3: Attendance check-in information of teachers and students in colleges and universities, uploading daily assessment score information and test papers of college students to conduct comprehensive evaluation of college teachers; S4: Evaluation encryption; S5: Output comments.

2. The method for evaluating higher education based on big data technology according to claim 1 is characterized by: The classification of grades includes the following steps: Based on course code: If the grade file contains a course code, the system can directly query the corresponding course in the database based on the course code and obtain the information of the main teacher of the course; Based on teacher ID: Some systems allow teacher IDs to be specified directly in the grade file, so that even if there are multiple teachers teaching the course, the grade can be accurately assigned to each teacher; Multi-condition matching: For complex scenarios, such as when a course is taught by multiple teachers in different classes, the system can achieve more accurate matching by combining multiple conditions such as course code and class number.

3. The method for evaluating higher education based on big data technology according to claim 1 is characterized by: The test difficulty assessment includes the following steps: Test paper and question import: import all the contents of the test paper and question into the system; Difficulty analysis of test questions: evaluate the difficulty of test questions; Comparison: Analyze the difficulty of the test questions based on the imported test questions, calculate the P value of each question (that is, the proportion of people who answered the question correctly) according to the candidates' answers, and add up the P values ​​of multiple questions in the test paper to calculate the average P value. If the average P value is significantly lower than expected, it means that the question is too difficult. The system determines that the score of the single subject is invalid, and the score is not included in the teacher's evaluation reference to avoid the accuracy of the teacher's evaluation being affected by the difficulty of the test questions that are not due to the teacher's teaching. On the contrary, when the P value is not lower than the preset value, the score is included in the teacher's evaluation reference.

4. The method for evaluating higher education based on big data technology according to claim 1 is characterized by: Evaluation encryption includes the following steps: Data extraction: extract teacher-related information from the database based on the corresponding teachers evaluated; Data encryption: Encrypt the evaluation data based on teacher-related information; Generate key: Encrypt the evaluation information and generate questions. After passing the questions, you can view the evaluation information of the relevant teachers.

5. The method for evaluating higher education based on big data technology according to claim 2 is characterized by: The course code-based approach uses either writing SQL queries or using queries in an ORM framework.

6. The method for evaluating higher education based on big data technology according to claim 3 is characterized by: The evaluation of the difficulty of the test questions adopts the item response theory (IRT) algorithm to evaluate the difficulty of the test questions.

7. The method for evaluating higher education based on big data technology according to claim 3 is characterized by: The difficulty comparison of the test questions is performed using a decision tree and a support vector machine (SVM).

8. The method for evaluating higher education based on big data technology according to claim 4 is characterized by: The data extraction uses natural language processing (NLP) to extract teacher-related information from the database based on the corresponding teachers evaluated.

9. The method for evaluating higher education based on big data technology according to claim 4 is characterized by: After the key is generated, a digital signature or date of birth can be generated based on the teacher's information to unlock and view the evaluation information.

10. A higher education evaluation system based on big data technology, characterized by: Used to execute the higher education evaluation method based on big data technology as described in any one of claims 1-9.

Images