Intelligent teaching effect evaluation system based on student behavior data

Abstract

The invention discloses an intelligent teaching effect evaluation system based on student behavior data, and relates to the technical field of teaching effect evaluation, and the system specifically comprises a behavior data extraction module, a course classification module and a teaching quality evaluation module, the course classification module is used for extracting teacher behavior characteristics according to classroom speaking duration, speed and blackboard-writing frequency of teachers, setting teaching indexes based on the teacher behavior characteristics and classifying offline behavior data through the teaching indexes, and the teaching quality evaluation module is used for evaluating teaching quality of the teachers. The system is used for integrating offline behavior data of different curriculum types, setting offline teaching evaluation indexes, setting online teaching evaluation indexes according to online behavior data and comprehensively evaluating the teaching quality of teachers, and intelligent teaching quality evaluation is achieved by integrating online and offline data.

Description

Technical Field

[0001] The present invention belongs to the technical field of teaching effect evaluation, and particularly relates to an intelligent teaching effect evaluation system based on student behavior data. Background Art

[0002] In modern education, teaching effect evaluation is an important link for teachers to improve teaching quality and ensure students' learning effects. Traditional evaluation methods mostly rely on single-dimensional data such as students' exam scores and homework completion, and these methods often cannot comprehensively and accurately reflect teachers' teaching effects and students' learning behaviors. With the development of information technology, especially the wide application of technologies such as big data, artificial intelligence, and the Internet of Things, the education field has gradually started to use multi-modal data to comprehensively evaluate teaching effects.

[0003] Multi-modal data refers to various types of data collected through multiple sensors and platforms that can reflect students' learning and teachers' teaching. These data include, but are not limited to, students' online learning behavior data (such as video viewing duration, quiz submission), and offline classroom behavior data (such as students' head orientation, activity frequency, teacher's blackboard writing frequency). By comprehensively analyzing these multi-modal data, a more comprehensive and in-depth understanding of students' learning status and teachers' teaching effects can be obtained. However, existing multi-modal data evaluation methods still have some technical problems, including difficulty in effectively collecting and integrating online and offline multi-modal behavior data, inability to accurately extract behavior characteristics and classify them, and how to set reasonable teaching effect evaluation indicators according to different course types.

[0004] Therefore, there is an urgent need for an intelligent teaching effect evaluation system based on student behavior data to solve the above problems. Summary of the Invention

[0005] The purpose of the present invention is to provide an intelligent teaching effect evaluation system based on student behavior data, which is used to solve the technical problems in the prior art, such as difficulty in integrating online and offline multi-modal behavior data, extracting behavior characteristics and classifying them, and setting reasonable teaching effect evaluation indicators according to different course types.

[0006] To achieve the above purpose, the present invention adopts the following technical solutions: An intelligent teaching effect evaluation system based on student behavior data, comprising: A behavior data extraction module, which is used to comprehensively collect multi-modal behavior data related to students' learning, associate the multi-modal behavior data and extract features; A course classification module, which is used to extract teachers' behavior characteristics according to the teachers' classroom speaking duration, speaking speed, and blackboard writing frequency, set teaching indicators based on the teachers' behavior characteristics, and classify the offline behavior data through the teaching indicators; The teaching quality evaluation module is used to integrate offline behavior data of different course types, set offline teaching evaluation indicators, set online teaching evaluation indicators based on online behavior data, and comprehensively evaluate the teaching quality of teachers.

[0007] Furthermore, the multimodal behavior data related to students' learning is collected. The specific method is as follows: According to the teaching objectives, the group of teachers participating in the evaluation is screened, and the fixed duration T is used as the data collection cycle to collect the behavioral data of students during the teaching process of this type of teachers. According to the differences in teaching platforms, the student behavioral data are divided into online behavioral data and offline behavioral data. The online learning data of each student is obtained through the learning tool interoperability interface of the online learning platform to obtain the online behavioral data. By connecting the classroom camera, the behavioral data of students and teachers in the classroom during the teaching period are obtained. OpenCV is used to identify the student's head direction, student activity frequency and teacher blackboard writing frequency during the teaching period. ASR is used to read the classroom voice content in the camera content in real time, including determining the speaking speed, decibel level of the student's classroom speech content, and the duration and speaking speed of the teacher's classroom speech. The student data collected in the classroom are recorded as offline behavioral data.

[0008] Furthermore, it is used to extract teacher behavior characteristics based on the teacher's classroom speaking time, speaking speed and blackboard writing frequency. The specific method is as follows: Extract the processed teacher's voice clips, identify the teacher's voiceprint in the voice clips through ASR, set the duration of the teacher's voiceprint in the voice clips greater than r as the total duration of the teacher's class speech, determine the total duration of the teacher's class speech in each class, and combine the class duration to obtain the proportion of the teacher's speech time in the class. The proportion of speech time is equal to the ratio of the total class speech time to the class duration; Use NLP tools to remove student speech and invalid noise segments from classroom voice segments, extract the speech text from the teacher's voice segments, count the total number of words in the teacher's speech text and the teacher's speech time, calculate the ratio of the total number of words in the teacher's speech text to the teacher's speech time in a class, and obtain the teacher's speech speed frequency in the class; Through OpenCV, the teacher's blackboard writing actions in the classroom are dynamically detected. The action of the teacher touching the blackboard with chalk is recorded as the blackboard writing start action, the action of the teacher touching the blackboard with a blackboard eraser is recorded as the blackboard writing update action, and the time interval from the teacher completing one blackboard writing content to the next update of the blackboard content is recorded as the blackboard writing interval. The ratio of the number of blackboard writing updates by the teacher in a class to the average blackboard writing interval is calculated to obtain the blackboard writing frequency. If the teacher does not update the blackboard writing in a class, the blackboard writing frequency of that class is 0.

[0009] Furthermore, the specific method of setting teaching indicators based on teacher behavior characteristics is as follows: Set teaching indicators based on the proportion of the teacher's speaking time in class, the teacher's speech rate frequency, and the blackboard writing rate, and use the formula Denoted by B(i), where i represents the i-th class, and Tu(i) represents the teaching indicator of the i-th class. Represents the proportion of the teacher's speaking time in the i-th class. Represents the weight coefficient of the speaking time proportion. Represents the speech rate frequency of the teacher in the i-th class. Represents the weight coefficient of the speech rate frequency. Represents the preset optimal speech rate frequency, and B(i) represents the blackboard writing rate of the teacher in the i-th class. Represents the weight coefficient of the blackboard writing rate.

[0010] Furthermore, classify the offline behavior data through the teaching indicators. The specific method is as follows: Preset the teaching indicator threshold A for teaching-type courses. When is greater than or equal to A, and satisfies is greater than or equal to a1, is greater than or equal to b1, or B(i) is greater than or equal to c1, it is determined that the i-th class is a teaching-type course. Preset the teaching indicator threshold B for self-study-type courses. When is less than or equal to B, and satisfies is less than or equal to a2, is greater than or equal to b2, or B(i) is greater than or equal to c2, it is determined that the i-th class is a self-study-type course. For courses where B < A, set a correction factor according to the number of questions asked by students in the student behavior data within the course. If + is greater than or equal to the preset threshold F, then the course is classified as a teaching-type course; otherwise, the course is classified as a self-study-type course.

[0011] Furthermore, set the correction factor , and the specific method is as follows: Use the formula Denotes the correction factor, where i represents the i-th class. Represents the correction factor of the i-th class. Represents the number of questions asked by students in the i-th class. Represents the class time of the i-th class, and m is a constant.

[0012] Furthermore, comprehensively consider the offline behavior data of different course types and set offline teaching evaluation indicators. The specific method is as follows: According to the offline course classification results, extract the video frames of teaching-type courses at a frame rate of G, and use the formula Indicates the offline teaching evaluation index. Among them, t represents the t-th data collection cycle, j represents the j-th video frame, N(t) represents the number of video frames of teaching courses extracted in the t-th data collection cycle, represents the number of students with their heads facing the blackboard, the teacher, or the desk in the j-th video frame, v represents the v-th teaching course, represents that in the t-th data collection cycle, it altogether contains teaching courses, represents the total number of speaking students in the v-th class, represents the total number of students in the v-th class, represents that in the t-th data collection cycle, it altogether contains self-study courses, g is a preset course proportion threshold, q represents the concentration weight coefficient, and p represents the positive weight coefficient.

[0013] Furthermore, set the online teaching evaluation index according to the online behavior data. The specific method is as follows: Define that when the duration of a student's single continuous video viewing is greater than or equal to O and the fast-forward operation is less than qr times, this duration is recorded as the student's effective learning duration. If the single-day viewing duration is greater than or equal to twice the total course duration and the proportion of the fast-forward operation is greater than or equal to tr%, mark this data as abnormal data and delete the records that have not been opened after the courseware is downloaded. Use the formula to represent the online teaching evaluation index. Among them, t represents the t-th data collection cycle, and y represents the y-th online video course, represents that in the t-th data collection cycle, it altogether contains online video courses, represents the total duration of the y-th online video course, represents the average effective learning duration of the student for the y-th online video course, represents the courseware weight coefficient corresponding to the y-th online video course, and x represents the x-th online test, represents that in the t-th data collection cycle, it altogether contains online tests, represents the average time taken by the student to complete the x-th online test, represents the preset standard completion duration of the x-th online test, represents the average number of times the student modifies the answers for the x-th online test, represents the total number of questions in the x-th online test, represents the effectiveness weight coefficient, c represents the conscientiousness weight coefficient, and f(t) represents the total number of abnormal data in the t data collection cycles.

[0014] Furthermore, comprehensively evaluate the teaching quality of teachers. The specific method is as follows: Set the offline teaching evaluation index threshold Fa and the online teaching evaluation index threshold Fb. Compare the calculated offline teaching evaluation index and online teaching evaluation index with the corresponding thresholds Fa and Fb. If the offline teaching evaluation index is greater than or equal to the threshold Fa and the online teaching evaluation index is greater than or equal to the threshold Fb, it indicates that the teaching effect of the teacher is good during this data collection period; If the offline teaching evaluation index is greater than or equal to the threshold Fa and the online teaching evaluation index is less than the threshold Fb, it indicates that the teacher's offline teaching effect is good and there are defects in online teaching during this data collection period; If the offline teaching evaluation index is less than the threshold Fa and the online teaching evaluation index is greater than or equal to the threshold Fb, it indicates that there are defects in the teacher's offline teaching level during this data collection period; If the offline teaching evaluation index is less than the threshold Fa and the online teaching evaluation index is less than the threshold Fb, it is necessary to indicate that the teacher's teaching quality problem is obvious during this data collection period.

[0015] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. By comprehensively collecting multi-modal behavior data related to students' learning, the present invention comprehensively reflects the learning status and behavior habits of students, which helps to improve the accuracy and reliability of evaluation. By analyzing teachers' behavior characteristics to set teaching indicators, it realizes the detailed classification of offline behavior data, helps to distinguish different types of courses, and helps to improve the scientificity of teaching quality evaluation; 2. By comprehensively integrating offline behavior data of different course types and setting corresponding evaluation indicators, the offline teaching evaluation index takes into account students' concentration and enthusiasm in class, and obtains relevant data through methods such as video frame analysis and number statistics, providing an intuitive and reliable basis for evaluation.

[0016] 3. By obtaining online behavior data through the learning tool interoperability interface of the online learning platform and comprehensively setting online teaching evaluation indicators, it pays attention to data such as students' effective learning duration and test performance. By defining concepts such as effective learning duration and abnormal data, it ensures the accuracy and effectiveness of evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0018] Figure 1Shows a module diagram of an intelligent teaching effect evaluation system based on student behavior data; Figure 2 Shows a method step diagram of an intelligent teaching effect evaluation method based on student behavior data; Figure 3 Shows a method step diagram of an offline course type judgment method. Specific implementation manners

[0019] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0020] Such as Figure 1 、 Figure 2 、 Figure 3 The intelligent teaching effect evaluation system based on student behavior data shown specifically includes the following: A behavior data extraction module comprehensively collects multi-modal behavior data related to students' learning, correlates the multi-modal behavior data, and performs feature extraction.

[0021] According to teaching objectives, such as subject type and class size, screen the teacher group participating in the evaluation. With a fixed time period T as the data collection cycle, collect the behavior data of students during the teaching process of this type of teacher, and divide the behavior data of students into online behavior data and offline behavior data according to the difference of the teaching platform. Among them, online behavior data refers to the behavior data of students when learning through an online learning platform, and offline behavior data refers to the behavior data generated by the interaction between teachers and students during offline classroom teaching; All acquisition devices and the server synchronize time using the NTP protocol, and the error is controlled within ±a milliseconds. In this embodiment, a is set to 10. The online learning platform includes course video content, teacher courseware content, and quiz answering content. Through the built-in interface of the platform, such as using the xAPI (Experience API) standard, capture students' learning behavior events such as "watching videos" and "submitting quizzes", and obtain the online learning data of each student through the learning tool interoperability interface of the online learning platform to obtain online behavior data. The online behavior data includes the playing duration of course video content, the frequency of courseware downloads, and quiz answering records; By connecting the classroom cameras, behavioral data of students and teachers in the classroom during teaching is obtained. Through OpenCV, the head orientation of students, the activity frequency of students, and the blackboard writing frequency of teachers during teaching are identified. Through ASR, the classroom speech content in the camera content is read in real time, including determining the speech rate, sound decibel level of students' classroom speech content, the duration and speech rate of teachers' classroom speech. The student data collected in the classroom is recorded as offline behavioral data.

[0022] The collected data is collectively referred to as teaching multi-modal data. A unified timestamp is added to the teaching multi-modal data. The offline data is mapped to the coordinates of different seats in the classroom by establishing an offline space coordinate system. Invalid data is removed, such as outliers caused during the camera device failure time. Missing values are interpolated based on time series, and the sliding window algorithm is used to smooth the mouse trajectory noise. The processed teaching multi-modal data is sorted in chronological order.

[0023] The course classification module extracts teachers' behavioral characteristics based on the duration, speech rate, and blackboard writing frequency of teachers' classroom speech, sets teaching indicators based on teachers' behavioral characteristics, and classifies the offline behavioral data through the teaching indicators.

[0024] Extract the processed teacher voice segments, identify the teacher's voiceprint in the voice segments through ASR (Automatic Speech Recognition). Set the duration with the proportion of the teacher's voiceprint in the voice segment greater than r as the total duration of the teacher's classroom speech. Determine the total classroom speech duration of the teacher in each class, and combine the classroom duration to obtain the proportion of the teacher's speaking duration in this class. The proportion of speaking duration is equal to the ratio of the total classroom speech duration to the classroom duration; Use NLP (Natural Language Processing) tools to remove students' speeches and invalid noise segments in the classroom voice segments, extract the speech texts of the teacher in the voice segments, count the total number of words and the speaking duration of the teacher's speech texts, and calculate the ratio of the total number of words of the teacher's speech texts to the speaking duration in a class to obtain the teacher's speech rate frequency in this class; Dynamically detect the teacher's blackboard writing actions in the classroom through OpenCV. Record the action of the teacher touching the blackboard with chalk as the start action of blackboard writing, record the action of the teacher touching the blackboard with the blackboard eraser as the update action of blackboard writing, record the time interval from when the teacher completes one blackboard writing content to the next update of the blackboard writing content as the blackboard writing interval, and calculate the ratio of the number of blackboard writing updates of the teacher in a class to the average value of the blackboard writing interval to obtain the blackboard writing frequency. If there is no blackboard writing update by the teacher in a class, the blackboard writing frequency of this class is 0; Comprehensively set teaching indicators based on the proportion of the teacher's speaking duration in the class, the teacher's speech rate frequency, and the blackboard writing frequency, and distinguish the classroom types through the teaching indicators. The specific calculation formula of the teaching indicators is as follows: B(i); Among them, \(i\) represents the \(i\)-th class, and \(Tu(i)\) represents the teaching index of the \(i\)-th class. represents the proportion of the teacher's speaking time in the \(i\)-th class. represents the weight coefficient of the speaking time proportion. represents the speaking speed frequency of the teacher in the \(i\)-th class. represents the weight coefficient of the speaking speed frequency. represents the preset optimal speaking speed frequency, and \(B(i)\) represents the blackboard writing frequency of the teacher in the \(i\)-th class. represents the weight coefficient of the blackboard writing frequency. In this embodiment, \(maxS\) is set to 170 words per minute. is equal to 0.6. is equal to 0.5. is equal to 0.4.

[0025] Preset the teaching index threshold \(A\) for teaching classes. When is greater than or equal to \(A\) and satisfies is greater than or equal to \(a1\), is greater than or equal to \(b1\), or \(B(i)\) is greater than or equal to \(c1\), it is determined that the \(i\)-th class is a teaching type course. Preset the teaching index threshold \(B\) for self-study classes. When is less than or equal to \(B\) and satisfies is less than or equal to \(a2\), is greater than or equal to \(b2\), or \(B(i)\) is greater than or equal to \(c2\), it is determined that the \(i\)-th class is a self-study type course. For courses where \(B \lt\) \(A\), a correction factor is set according to the number of student questions in the student behavior data within the course, and the course type is judged by integrating the teaching index of the corresponding course. The specific calculation formula of the correction factor is as follows: ; Among them, \(i\) represents the \(i\)-th class. represents the correction factor of the \(i\)-th class. represents the number of student questions in the \(i\)-th class. represents the class time of the \(i\)-th class, and \(m\) is a constant value set based on the course duration. In this embodiment, \(m\) is set to 0.55; Courses with a speaking time proportion greater than or equal to \(B\) are screened, and a threshold \(F\) is preset based on the average value of the sum of the teaching index and the correction factor of this type of course. In this embodiment, \(B\) is set to 37%. If +\( is greater than or equal to the preset threshold \(F\), then the course is classified as a teaching type course; otherwise, the course is classified as a self-study type course.

[0026] The teaching quality evaluation module integrates the offline behavior data of different course types, sets offline teaching evaluation indicators, sets online teaching evaluation indicators according to the online behavior data, and comprehensively evaluates the teaching quality of teachers.

[0027] According to the classification results of offline courses, define differentiated evaluation dimensions, extract video frames of teaching courses at G frame rate, and set offline teaching evaluation indicators by integrating students' offline behavior data. The specific formula is as follows: ; Among them, t represents the t-th data collection period, j represents the j-th video frame, N(t) represents the total number of video frames of teaching courses extracted in the t-th data collection period, represents the number of students with their heads facing the blackboard, the teacher or the desk in the j-th video frame, v represents the v-th teaching course, represents that the t-th data collection period contains a total of teaching courses, represents the total number of speaking students in the v-th class, represents the total number of students in the v-th class, represents that the t-th data collection period contains a total of self-study courses, g is a preset course proportion threshold, q represents the focus weight coefficient, and p represents the positive weight coefficient.

[0028] Define that when the duration of a student's single continuous video viewing is greater than or equal to O and the fast-forward operation is less than qr times, this duration is recorded as the student's effective learning duration. If the single-day viewing duration is greater than or equal to twice the total course duration and the proportion of fast-forward operations is greater than or equal to tr%, this data is marked as abnormal data, and records that have not been opened after the courseware is downloaded are deleted. In this embodiment, O is set to 30 seconds, qr is set to 2, and tr is set to 80. Set online teaching evaluation indicators by integrating students' online behavior data. The specific formula is as follows: ; Among them, t represents the t-th data collection period, y represents the y-th online video course, represents that the t-th data collection period contains a total of online video courses, represents the total duration of the y-th online video course, represents the average effective learning duration of students learning the y-th online video course, It represents the courseware weight coefficient corresponding to the y-th online video course. In this embodiment, it is set that when more than 50% of the students download the courseware before watching the course, the corresponding courseware weight coefficient is equal to 1.2. If more than 50% of the students download the courseware during the process of watching the course, the corresponding courseware weight coefficient is equal to 1. In other cases, the corresponding courseware weight is equal to 0.8. x represents the x-th online test. It represents that there are a total of times of online tests within the t-th data collection period. It represents the average time taken by students to complete the x-th online test. It represents the preset standard completion duration of the x-th online test. It represents the average number of times students modify their answers for the x-th online test. It represents the total number of questions in the x-th online test. It represents the effectiveness weight coefficient, c represents the conscientiousness weight coefficient, and f(t) represents the total number of abnormal data within the t data collection periods.

[0029] Based on the data averages of the online teaching evaluation indicators and the online teaching evaluation indicators over a past period of time, the offline teaching evaluation indicator threshold Fa and the online teaching evaluation indicator threshold Fb are respectively set. The calculated offline teaching evaluation indicator, online teaching evaluation indicator are compared with the corresponding thresholds Fa and Fb. If the offline teaching evaluation indicator is greater than or equal to the threshold Fa and the online teaching evaluation indicator is greater than or equal to the threshold Fb, it indicates that the teaching effect of the teacher is good during this data collection period and the teaching method needs to be continued. If the offline teaching evaluation indicator is greater than or equal to the threshold Fa and the online teaching evaluation indicator is less than the threshold Fb, it indicates that the teacher's offline teaching effect is good during this data collection period and there are defects in the online teaching, and it is necessary to improve the online teaching method, including adjusting the test difficulty and modifying the duration of the online course video. If the offline teaching evaluation indicator is less than the threshold Fa and the online teaching evaluation indicator is greater than or equal to the threshold Fb, it indicates that the teacher needs to improve the offline teaching level and improve the offline teaching method during this data collection period. For example: improving the content of the blackboard writing and adjusting the proportion of different course types. If the offline teaching evaluation indicator is less than the threshold Fa and the online teaching evaluation indicator is less than the threshold Fb, it indicates that the teacher's teaching quality problem is obvious during this data collection period and a teaching warning needs to be given to this teacher.

[0030] As described above, it is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent replacements or changes, and all should be covered by the protection scope of the present invention.

[0031] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments only. Obviously, according to the content of this specification, many modifications and changes can be made. These embodiments are selected and specifically described in this specification in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and utilize the present invention well. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. An intelligent teaching effect evaluation system based on student behavior data, characterized in that, include: The behavior data extraction module is used to comprehensively collect students' learning-related multimodal behavior data, associate the multimodal behavior data and perform feature extraction; The course classification module is used to extract teacher behavior characteristics based on the teacher's classroom speaking time, speaking speed and blackboard writing frequency, set teaching indicators based on the teacher's behavior characteristics, and classify offline behavior data through teaching indicators; The teaching quality evaluation module is used to integrate offline behavior data of different course types, set offline teaching evaluation indicators, set online teaching evaluation indicators based on online behavior data, and comprehensively evaluate the teaching quality of teachers.

2. The intelligent teaching effect evaluation system based on student behavior data according to claim 1, characterized in that, Collect students' learning-related multimodal behavior data. The specific methods are as follows: According to the teaching objectives, the group of teachers participating in the evaluation are screened, and the fixed duration T is used as the data collection period. During the teaching process of the group of teachers participating in the evaluation, student behavior data are collected. According to the differences in teaching platforms, the student behavior data are divided into online behavior data and offline behavior data. The online learning data of each student is obtained through the learning tool interoperability interface of the online learning platform to obtain online behavior data. By connecting to the classroom camera, the behavior data of students and teachers in the classroom during teaching are obtained. OpenCV is used to identify the student's head direction, student activity frequency and teacher blackboard writing frequency during teaching. ASR is used to read the classroom voice content in the camera content in real time, including determining the speaking speed and decibel level of the student's classroom speech content, and the duration and speaking speed of the teacher's classroom speech. The student data collected in the classroom are recorded as offline behavior data.

3. The intelligent teaching effect evaluation system based on student behavior data according to claim 1, characterized in that It is used to extract teacher behavior characteristics based on the teacher's classroom speaking time, speaking speed and blackboard writing frequency. The specific method is: Extract the processed teacher's voice clips, identify the teacher's voiceprint in the voice clips through ASR, set the duration of the teacher's voiceprint in the voice clips greater than r as the total duration of the teacher's class speech, determine the total duration of the teacher's class speech in each class, and combine the class duration to obtain the proportion of the teacher's speech time in the class. The proportion of speech time is equal to the ratio of the total class speech time to the class duration; Use NLP tools to remove student speech and invalid noise segments from classroom voice segments, extract the speech text from the teacher's voice segments, count the total number of words in the teacher's speech text and the teacher's speech time, calculate the ratio of the total number of words in the teacher's speech text to the teacher's speech time in a class, and obtain the teacher's speech speed frequency in the class; Through OpenCV, the teacher's blackboard writing actions in the classroom are dynamically detected. The action of the teacher touching the blackboard with chalk is recorded as the blackboard writing start action, the action of the teacher touching the blackboard with a blackboard eraser is recorded as the blackboard writing update action, and the time interval from the teacher completing one blackboard writing content to the next update of the blackboard content is recorded as the blackboard writing interval. The ratio of the number of blackboard writing updates by the teacher in a class to the average blackboard writing interval is calculated to obtain the blackboard writing frequency. If the teacher does not update the blackboard writing in a class, the blackboard writing frequency of that class is 0.

4. The intelligent teaching effect evaluation system based on student behavior data according to claim 3, wherein, The specific method of setting teaching indicators based on teacher behavior characteristics is: Set teaching indicators based on the proportion of the teacher's speaking time in class, the teacher's speech rate frequency, and the blackboard writing frequency, and use the formula Denoted by B(i), where i represents the i-th class, and Tu(i) represents the teaching indicator of the i-th class. Represents the proportion of the teacher's speaking time in the i-th class. Represents the weight coefficient of the speaking time proportion. Represents the speech rate frequency of the teacher in the i-th class. Represents the weight coefficient of the speech rate frequency. Represents the preset optimal speech rate frequency, and B(i) represents the blackboard writing frequency of the teacher in the i-th class. Represents the weight coefficient of the blackboard writing frequency.

5. The intelligent teaching effect evaluation system based on student behavior data according to claim 4, wherein The offline behavior data is classified by teaching indicators. The specific method is as follows: Preset the teaching index threshold A for teaching courses. When is greater than or equal to A and satisfies is greater than or equal to a1, is greater than or equal to b1, or B(i) is greater than or equal to c1, determine that the i-th class is a teaching type course; Preset the self-study type teaching index threshold B. When is less than or equal to B and satisfies is less than or equal to a2, is greater than or equal to b2, and B(i) is greater than or equal to c2, determine that the i-th class is a self-study type course; For course B< of A, a correction factor is set according to the number of questions asked by students in the student behavior data within the course . If + is greater than or equal to the preset threshold F, then the course is classified as a teaching type course; otherwise, the course is classified as a self-study type course.

6. The intelligent teaching effect evaluation system based on student behavior data according to claim 5, wherein, Set correction factor , and the specific method is as follows: Using the formula represents the correction factor, where i represents the i-th class, represents the correction factor of the i-th class, represents the number of student questions in the i-th class, represents the class time of the i-th class, and m is a constant.

7. The intelligent teaching effect evaluation system based on student behavior data according to claim 1, characterized in that, Based on offline behavior data of different course types, offline teaching evaluation indicators are set. The specific method is as follows: According to the offline course classification results, video frames of teaching courses are extracted at G frame rate, and using the formula represents the offline teaching evaluation index, where t represents the t-th data collection period, j represents the j-th video frame, N(t) represents the total number of video frames of teaching courses extracted within the t-th data collection period, represents the number of students with their heads facing the blackboard, the teacher or the desk within the j-th video frame, v represents the v-th teaching course, represents that within the t-th data collection period, there are a total of teaching courses, represents the total number of speaking students in the v-th class, represents the total number of students in the v-th class, represents that within the t-th data collection period, there are a total of self-study courses, g is the preset course proportion threshold, q represents the concentration weight coefficient, and p represents the positive weight coefficient.

8. The intelligent teaching effect evaluation system based on student behavior data according to claim 1, characterized in that, Set online teaching evaluation indicators based on online behavior data. The specific method is as follows: When the continuous viewing duration of a student for a single video is greater than or equal to O and the number of fast-forward operations is less than qr times, this duration is recorded as the student's effective learning duration. If the single-day viewing duration is greater than or equal to twice the total course duration and the proportion of fast-forward operations is greater than or equal to tr%, this data is marked as abnormal data. Records of unopened courseware downloads are deleted. Using the formula represents the online teaching evaluation index. Among them, t represents the t-th data collection period, and y represents the y-th online video course. represents that there are a total of online video courses in the t-th data collection period. represents the total duration of the y-th online video course. represents the average effective learning duration of the student for the y-th online video course. represents the courseware weight coefficient corresponding to the y-th online video course. x represents the x-th online test. represents that there are a total of online tests in the t-th data collection period. represents the average time taken by the student to complete the x-th online test. represents the preset standard completion duration of the x-th online test. represents the average number of times the student modifies the answers for the x-th online test. represents the total number of questions in the x-th online test. represents the validity weight coefficient, c represents the conscientiousness weight coefficient, and f(t) represents the total number of abnormal data in the t data collection periods.

9. The intelligent teaching effect evaluation system based on student behavior data according to claim 1, characterized in that Comprehensively evaluate the teaching quality of teachers. The specific method is as follows: Set the threshold Fa for offline teaching evaluation indicators and the threshold Fb for online teaching evaluation indicators. Compare the calculated offline teaching evaluation indicators, online teaching evaluation indicators with the corresponding thresholds Fa and Fb. If the offline teaching evaluation indicator is greater than or equal to the threshold Fa and the online teaching evaluation indicator is greater than or equal to the threshold Fb, it indicates that the teaching effect of the teacher is good during this data collection period; If the offline teaching evaluation indicator is greater than or equal to the threshold Fa and the online teaching evaluation indicator is less than the threshold Fb, it indicates that the teacher's offline teaching effect is good and there are defects in online teaching during this data collection period; If the offline teaching evaluation indicator is less than the threshold Fa and the online teaching evaluation indicator is greater than or equal to the threshold Fb, it indicates that the teacher's offline teaching level has defects during this data collection period; If the offline teaching evaluation indicator is less than the threshold Fa and the online teaching evaluation indicator is less than the threshold Fb, it is necessary to indicate that the teacher's teaching quality problem is obvious during this data collection period.

Images