Teaching level evaluation method and system based on big data

Abstract

The invention relates to the technical field of intelligent assessment, and particularly discloses a teaching level assessment method and system based on big data, and the method comprises the steps: recording classroom interaction information based on an intelligent device, recognizing the classroom interaction information, and extracting dialogue information; acquiring a student score array containing time nodes, and performing arrangement according to a time sequence to obtain a student score matrix; comparing dialogue information of different classes, clustering the classes, reading a student score matrix of the same class, and performing internal correction on the student score matrix of the same class; and receiving scoring information input by the students, and counting the scoring information and the corrected student score matrix to determine a total evaluation score. According to the method, the student score matrix is corrected according to the dialogue information to serve as the evaluation parameter, the objectivity is extremely high, the existing subjective evaluation process is combined, subjective and objective combination is achieved, and the evaluation process is enriched.

Description

Technical Field

[0001] The present invention relates to the field of intelligent assessment technology, and in particular to a teaching level assessment method and system based on big data. Background Art

[0002] Teaching level assessment is a process of systematically evaluating the teaching ability, teaching effect and professional quality demonstrated by teachers in the teaching process. The existing teaching level assessment method is relatively simple, that is, after the teaching process is completed, the students will score it and then get the assessment result. This method is too subjective and not objective enough. In fact, with the popularization of smart devices in the classroom, the evaluation process of the teaching process can also introduce a more objective evaluation scheme. How to objectively evaluate the teaching process with the help of smart devices is the technical problem that the technical solution of the present invention wants to solve. Summary of the invention

[0003] The purpose of the present invention is to provide a teaching level evaluation method and system based on big data to solve the problems raised in the above background technology.

[0004] To achieve the above object, the present invention provides the following technical solutions: A teaching level evaluation method based on big data, the method comprising: Based on recording classroom interaction information by a smart device, the classroom interaction information is identified to extract dialogue information; wherein the classroom interaction information at least includes audio information; the classroom interaction information and the extracted dialogue information have the same time period label; Obtain the student score array containing the time nodes, arrange them in chronological order, and obtain a student score matrix; the rows of the student score matrix correspond to the time nodes, the columns correspond to the students, and the element values ​​in the matrix represent the scores; Compare the conversation information of different classes, cluster the classes, read the student score matrix of the same class, and make internal corrections to the student score matrix of the same class; the internal correction method is to reduce the difference; Receive the scoring information entered by the students, and determine the total evaluation score by combining the scoring information and the revised student score matrix.

[0005] As a further solution of the present invention: the step of recording classroom interaction information based on a smart device, identifying the classroom interaction information, and extracting dialogue information includes: Real-time recording of classroom interaction information based on smart devices; When the classroom interaction information is audio, identifying the audio features of the teacher according to preset keywords; The audio is segmented according to the audio features to obtain the teaching audio and the non-teaching audio; wherein the teaching audio and the non-teaching audio both contain a timestamp; Audio recognition is performed on the audio of the teacher and the audio of the non-teacher to obtain dialogue information.

[0006] As a further solution of the present invention, the steps of comparing the conversation information of different classes, clustering the classes, reading the student score matrix of the same class, and internally correcting the student score matrix of the same class include: For any classroom conversation information, perform word recognition on it, and simplify the conversation information into a sequence of key words according to the word recognition results; each key word group corresponds to a conversation; Input the sequence of keyword groups into the preset conversion model to obtain a sequence of word vector groups; Compare sequences of word vector groups of different classes, calculate similarities, and cluster the classes based on the similarities; Read the student score matrix of similar classes and make internal corrections to the student score matrix of similar classes.

[0007] As a further solution of the present invention: the step of receiving the scoring information input by the students, and determining the total evaluation score by counting the scoring information and the modified student score matrix comprises: Receive scoring information input by students; The scoring information and the modified student score matrix are input into a trained evaluation model to obtain a total evaluation score; wherein the evaluation model is a preset mathematical statistical function.

[0008] As a further solution of the present invention: the method further comprises: For the student score matrix of any class, calculate the mean and standard deviation of each row of data to obtain the mean column and standard deviation column; Compare the mean columns of different classes and the standard deviation columns of different classes, calculate the similarity of different classes, and cluster the classes based on the similarity; For any type of classroom, read the conversation information of each classroom, compare the conversation information, extract the majority value conversation information as the standard conversation information of this type of classroom; The majority-valued dialogue information is used to characterize the majority-valued features of dialogue information in the same class.

[0009] As a further solution of the present invention: the method further comprises: Receive the acquired conversation information uploaded by a certain class; Regularly compare and match the acquired conversation information with the existing conversation information to obtain matching classes; Query the student score matrix of the matching classroom and select the optimal student score matrix; The standard dialogue information corresponding to the optimal student score matrix is ​​read as reference data and fed back to the classroom that uploaded the dialogue information.

[0010] The technical solution of the present invention also provides a teaching level evaluation system based on big data, the system comprising: A dialogue information extraction module is used to record classroom interaction information based on a smart device, identify the classroom interaction information, and extract dialogue information; wherein the classroom interaction information at least includes audio information; the classroom interaction information and the extracted dialogue information have the same time period label; A matrix generation module is used to obtain a student score array containing time nodes, arrange them in chronological order, and obtain a student score matrix; the rows of the student score matrix correspond to time nodes, the columns correspond to students, and the element values ​​in the matrix represent scores; The matrix correction module is used to compare the conversation information of different classes, cluster the classes, read the student score matrix of the same class, and make internal corrections to the student score matrix of the same class; wherein the internal correction method is to reduce the difference; The evaluation total score generation module is used to receive the scoring information input by the students, and to determine the evaluation total score by counting the scoring information and the revised student score matrix.

[0011] As a further solution of the present invention: the dialogue information extraction module includes: An information recording unit, used to record classroom interaction information in real time based on smart devices; An audio feature extraction unit, used for identifying the audio features of the teacher according to preset keywords when the classroom interaction information is audio; An audio segmentation unit is used to segment the audio according to the audio features to obtain the teaching audio and the non-teaching audio; wherein the teaching audio and the non-teaching audio both contain a timestamp; The audio recognition unit is used to perform audio recognition on the audio of the teacher and the audio of the non-teacher to obtain dialogue information.

[0012] As a further solution of the present invention: the matrix correction module includes: The information conversion unit is used to perform word recognition on the dialogue information of any class, and simplify the dialogue information into a sequence of key words according to the word recognition result; each key word group corresponds to a dialogue; A word conversion unit, used to input the sequence of keyword groups into a preset conversion model to obtain a sequence of word vector groups; A comparison and clustering unit, used to compare sequences of word vector groups of different classes, calculate similarities, and cluster the classes based on the similarities; The correction execution unit is used to read the student score matrix of the same class and perform internal correction on the student score matrix of the same class.

[0013] As a further solution of the present invention: the evaluation total score generating module includes: A scoring information receiving unit, used to receive scoring information input by students; The evaluation execution unit is used to input the scoring information and the modified student score matrix into the trained evaluation model to obtain the total evaluation score; wherein the evaluation model is a preset mathematical statistical function.

[0014] Compared with the prior art, the beneficial effects of the present invention are: the present invention records classroom interaction information based on smart devices, determines dialogue information based on the classroom interaction information, synchronously obtains the scores of students at each time, obtains the student score matrix, and corrects the student score matrix based on the dialogue information. As an evaluation parameter, the present invention is extremely objective, and combined with the existing subjective evaluation process, it combines subjective and objective, enriching the evaluation process. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention.

[0016] Figure 1 This is a flowchart of the teaching level evaluation method based on big data.

[0017] Figure 2 This is a structural diagram of the teaching level evaluation system based on big data. DETAILED DESCRIPTION

[0018] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0019] Figure 1 The flowchart of the teaching level evaluation method based on big data is shown in FIG. 1 . In an embodiment of the present invention, a teaching level evaluation method based on big data includes: Step S100: Based on the classroom interaction information recorded by the smart device, the classroom interaction information is identified and the dialogue information is extracted; wherein the classroom interaction information at least includes audio information; the classroom interaction information and the extracted dialogue information have the same time period label; The technical solution of the present invention is applied to smart classrooms. Smart classrooms are classrooms equipped with smart devices, which are very common in existing teaching environments. Many students are equipped with tablets during class, and the process of recording interactive information is very simple. In the technical solution of the present invention, the recorded interactive information is limited to audio information, and generally does not involve image information. On the one hand, the amount of image information is large, and on the other hand, everyone may be more resistant to image information and will not easily grant authorization. The classroom interactive information is identified to extract dialogue information.

[0020] It should be noted that even to obtain audio information, it is necessary to obtain the permissions granted by the teacher and the students in advance. If you do not have the permissions, you cannot perform subsequent processing.

[0021] Step S200: Obtain a student score array containing time nodes, and arrange them in chronological order to obtain a student score matrix; the rows of the student score matrix correspond to time nodes, the columns correspond to students, and the element values ​​in the matrix represent scores; Get the student scores for each exam, count them according to the same serial number, and get the student score array for each time node; sort them in chronological order, and the student score array can be spliced ​​into a matrix, called the student score matrix.

[0022] It should be noted that the data acquisition processes of step S100 and step S200 are both final statistical processes after a teaching process is completed. For example, a teaching process has 36 classes in total, the classroom interaction information is all the classroom interaction information within these 36 classes, and the student score matrix is ​​the scores of all exams within these 36 classes.

[0023] Step S300: Compare the conversation information of different classes, cluster the classes, read the student score matrix of the same class, and make internal corrections to the student score matrix of the same class; wherein the internal correction is made in a way of reducing the difference; The data statistics process of step S100 and step S200 is performed on multiple classes, which is itself a big data acquisition architecture. The corresponding number of classes is extremely large. By comparing the conversation information of different classes, the classes can be clustered according to the comparison results, and classes with similar teaching processes are classified into one category. Then, the student score matrix of the same class is read, and the student score matrix of the same class is internally corrected. The goal of the internal correction is to reduce the difference between the best student score matrix and the worst student score matrix. The purpose of reducing the difference is to reduce the difference in teaching effects caused by student differences. In fact, it is also feasible to calculate the mean of the student score matrix of the same class as the final evaluation of its teaching results, but this is too balanced and erases the differences between teachers, including differences in details such as facial expressions, demeanor and standing positions. Therefore, this application only narrows the differences, but does not completely erase the differences.

[0024] Step S400: receiving the scoring information input by the students, and calculating the scoring information and the corrected student score matrix to determine the total evaluation score; The existing evaluation system is that students give anonymous scores, which is highly subjective and easily interfered with, but it is a necessary scoring mechanism, so the present invention retains it; in the technical solution of the present invention, the generated student score matrix actually represents objective information, combining subjective and objective aspects to evaluate the teaching process more accurately.

[0025] Regarding step S100, the step of recording classroom interaction information based on a smart device, identifying the classroom interaction information, and extracting dialogue information includes: Real-time recording of classroom interaction information based on smart devices; When the classroom interaction information is audio, identifying the audio features of the teacher according to preset keywords; The audio is segmented according to the audio features to obtain the teaching audio and the non-teaching audio; wherein the teaching audio and the non-teaching audio both contain a timestamp; Audio recognition is performed on the audio of the teacher and the audio of the non-teacher to obtain dialogue information.

[0026] The above content provides a specific solution for extracting conversation information, based on real-time recording of classroom interaction information on smart devices. When the classroom interaction information is audio, the audio features of the teacher are identified based on preset keywords. There are many keywords, such as "classmates". This is the audio often emitted by the teacher. It is identified, the frequency part is extracted, and then the audio features are extracted (equivalent to identifying the teacher's timbre). The audio (classroom interaction information) is segmented according to the audio features to obtain the teacher's audio and non-teacher's audio. Finally, audio recognition is performed on the teacher's audio and non-teacher's audio to obtain conversation information. There are many existing recognition schemes used in the audio recognition scheme. It is a simple audio-to-text process and will not be repeated here.

[0027] It is worth mentioning that this application does not identify students, but only teachers, and the rest are collectively referred to as non-teachers; in addition, both the teacher audio and the non-teachers audio contain timestamps to determine the order of the recognized text.

[0028] Regarding step S300, the steps of comparing the conversation information of different classes, clustering the classes, reading the student score matrix of the same class, and internally correcting the student score matrix of the same class include: For any classroom conversation information, perform word recognition on it, and simplify the conversation information into a sequence of key words according to the word recognition results; each key word group corresponds to a conversation; Input the sequence of keyword groups into the preset conversion model to obtain a sequence of word vector groups; Compare sequences of word vector groups of different classes, calculate similarities, and cluster the classes based on the similarities; Read the score matrix of students in similar classes and make internal corrections to the score matrix of students in similar classes; In an example of the technical solution of the present invention, the conversation information of each class is identified in turn, and the word recognition is performed on the conversation information of any class, and the conversation information is converted into a set of words. Each sentence in the conversation information can be converted into a keyword group. Therefore, the entire conversation information can be converted into a sequence of keyword groups, and the order in the sequence is the time order determined by the timestamp.

[0029] Then, the sequence of keyword groups is input into a preset conversion model, each word is converted into a word vector, the character string is quantized, and a sequence of word vector groups is obtained; the sequences of word vector groups of different classes are compared, the similarity is calculated, and the classes are clustered based on the similarity to obtain multiple categories of classes; the clustering scheme can adopt a K-means clustering scheme, the K value can be a preset value, which is generally not unique, and multiple K values ​​can be applied, multiple clustering can be performed, and then the best K value is selected.

[0030] Finally, for each type of class, read the student score matrix of each class, and modify any student score matrix according to the student score matrix of the same type of class to obtain the final student score matrix.

[0031] Wherein, the internal correction method is: Calculate the mean of each element in the student score matrix; For any element position in any student score matrix, calculate the mean of the element value and the corresponding mean as the corrected data; Among them, when there is a student score matrix at a certain element position that does not contain a corresponding value, the original value is retained during the internal correction process.

[0032] The above content provides a specific internal correction scheme. First, all the student score matrices are counted and the mean of the values ​​at each element position is calculated. Then, for any value in the student score matrix, the mean of it and the corresponding mean is calculated as the corrected value. The corrected value is closer to the mean. In fact, when correcting, a weight can be added. If you want the original value to have a greater impact, you can make its weight two-thirds. If you want the original value to have a smaller impact, you can make its weight one-third.

[0033] Finally, it should be noted that since the sizes of the student score matrices may be different, there may be only some student score matrices with values ​​at some positions. This application does not calculate such values ​​and directly retains the original values.

[0034] Regarding step S400, the step of receiving the scoring information input by the student and determining the total evaluation score by counting the scoring information and the revised student score matrix includes: Receive scoring information input by students; The scoring information and the modified student score matrix are input into a trained evaluation model to obtain a total evaluation score; wherein the evaluation model is a preset mathematical statistical function.

[0035] In one example of the technical solution of the present invention, the scoring information and the corrected student score matrix are input into a trained evaluation model to obtain a total evaluation score. The evaluation model is a preset mathematical statistical function, which is determined by the staff according to the specific situation. For example, the mean of the scoring information is calculated, and the average progress score of the students is calculated according to the student score matrix (the difference between the maximum and minimum values ​​of any column, and then the mean of the differences of all columns is calculated), and the mean and the average progress score are input into a preset binary linear function to calculate the total score.

[0036] As a preferred embodiment of the technical solution of the present invention, the method further includes: For the student score matrix of any class, calculate the mean and standard deviation of each row of data to obtain the mean column and standard deviation column; Compare the mean columns of different classes and the standard deviation columns of different classes, calculate the similarity of different classes, and cluster the classes based on the similarity; For any type of class, the conversation information of each class is read, the conversation information is compared, and the majority-valued conversation information is extracted as the standard conversation information of this type of class.

[0037] For the student score matrix of any class, feature extraction is performed to obtain the mean column and standard deviation column, the mean columns of different classes and the standard deviation columns of different classes are compared, and the similarity of different classes is calculated (with the help of the similarity calculation scheme of the array), and the classes are clustered based on the similarity, which can group classes with the same score changes into one category. On this basis, the same class is clustered for the second time. The secondary clustering scheme is to cluster according to the dialogue information, and it can be obtained what the dialogue information is like under the same score change. The class with the largest number of classes is selected, and its dialogue information is used as the standard dialogue information as the dependent variable of the student score matrix; the practical significance of this is that if teaching is carried out according to the standard dialogue information, the corresponding student score matrix can be obtained with a high probability.

[0038] In summary, the majority-valued dialogue information is used to characterize the majority-valued features of dialogue information in the same class.

[0039] As a preferred embodiment of the technical solution of the present invention, the method further includes: Receive the acquired conversation information uploaded by a certain class; Regularly compare and match the acquired conversation information with the existing conversation information to obtain matching classes; Query the student score matrix of the matching classroom and select the optimal student score matrix; The standard dialogue information corresponding to the optimal student score matrix is ​​read as reference data and fed back to the classroom that uploaded the dialogue information.

[0040] In the technical solution of the present invention, the teaching level evaluation process is performed after the teaching is completed and the final state is presented, so as to intelligently score the teacher. In fact, this information can also be used as a reference in the teaching process, and the acquired dialogue information uploaded by a certain classroom is received, and the acquired dialogue information is regularly compared and matched with the existing dialogue information (the dialogue information of the same length is intercepted from the existing dialogue information, and the similarity is calculated. When it is high enough, it is regarded as a successful match) to obtain the matched classroom. The timing may mean once every two class hours; there may be many matched classrooms, and the student score matrix of the matched classroom is queried, and the optimal student score matrix (with the highest average improvement score) is selected, which is regarded as the best state achieved in the existing data under the current situation. Then, on the basis of the standard dialogue information corresponding to each student score matrix, the standard dialogue information corresponding to the optimal student score matrix is ​​queried as a teaching reference, and fed back to the classroom that uploaded the dialogue information as a guide to assist its teaching process.

[0041] Figure 2 : is a composition structure diagram of a teaching level evaluation system based on big data. In an embodiment of the present invention, a teaching level evaluation system based on big data, the system 10 includes: The dialogue information extraction module 11 is used to record classroom interaction information based on a smart device, identify the classroom interaction information, and extract dialogue information; wherein the classroom interaction information at least includes audio information; the classroom interaction information and the extracted dialogue information have the same time period label; The matrix generation module 12 is used to obtain the student score array containing the time nodes, arrange them in chronological order, and obtain the student score matrix; the rows of the student score matrix correspond to the time nodes, the columns correspond to the students, and the element values ​​in the matrix represent the scores; The matrix correction module 13 is used to compare the conversation information of different classes, cluster the classes, read the student score matrix of the same class, and perform internal correction on the student score matrix of the same class; wherein the internal correction method is to reduce the difference; The evaluation total score generating module 14 is used to receive the scoring information input by the students, and to determine the evaluation total score by counting the scoring information and the revised student score matrix.

[0042] Furthermore, the dialogue information extraction module 11 includes: An information recording unit, used to record classroom interaction information in real time based on smart devices; An audio feature extraction unit, used for identifying the audio features of the teacher according to preset keywords when the classroom interaction information is audio; An audio segmentation unit is used to segment the audio according to the audio features to obtain the teaching audio and the non-teaching audio; wherein the teaching audio and the non-teaching audio both contain a timestamp; The audio recognition unit is used to perform audio recognition on the audio of the teacher and the audio of the non-teacher to obtain dialogue information.

[0043] Specifically, the matrix correction module 12 includes: The information conversion unit is used to perform word recognition on the dialogue information of any class, and simplify the dialogue information into a sequence of key words according to the word recognition result; each key word group corresponds to a dialogue; A word conversion unit, used to input the sequence of keyword groups into a preset conversion model to obtain a sequence of word vector groups; A comparison and clustering unit, used to compare sequences of word vector groups of different classes, calculate similarities, and cluster the classes based on the similarities; The correction execution unit is used to read the student score matrix of the same class and perform internal correction on the student score matrix of the same class.

[0044] In addition, the evaluation total score generating module 14 includes: A scoring information receiving unit, used to receive scoring information input by students; The evaluation execution unit is used to input the scoring information and the modified student score matrix into the trained evaluation model to obtain the total evaluation score; wherein the evaluation model is a preset mathematical statistical function.

[0045] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A teaching level evaluation method based on big data, characterized in that: The method comprises: Based on recording classroom interaction information by a smart device, the classroom interaction information is identified to extract dialogue information; wherein the classroom interaction information at least includes audio information; the classroom interaction information and the extracted dialogue information have the same time period label; Obtain the student score array containing the time nodes, arrange them in chronological order, and obtain a student score matrix; the rows of the student score matrix correspond to the time nodes, the columns correspond to the students, and the element values ​​in the matrix represent the scores; Compare the conversation information of different classes, cluster the classes, read the student score matrix of the same class, and make internal corrections to the student score matrix of the same class; the internal correction method is to reduce the difference; Receive the scoring information entered by the students, and determine the total evaluation score by combining the scoring information and the revised student score matrix.

2. The teaching level evaluation method based on big data according to claim 1 is characterized in that: The step of recording classroom interaction information based on a smart device, identifying the classroom interaction information, and extracting dialogue information comprises: Real-time recording of classroom interaction information based on smart devices; When the classroom interaction information is audio, identifying the audio features of the teacher according to preset keywords; The audio is segmented according to the audio features to obtain the teaching audio and the non-teaching audio; wherein the teaching audio and the non-teaching audio both contain a timestamp; Audio recognition is performed on the audio of the teacher and the audio of the non-teacher to obtain dialogue information.

3. The teaching level evaluation method based on big data according to claim 1 is characterized in that: The steps of comparing the conversation information of different classes, clustering the classes, reading the student score matrix of the same class, and internally correcting the student score matrix of the same class include: For any classroom conversation information, perform word recognition on it, and simplify the conversation information into a sequence of key words according to the word recognition results; each key word group corresponds to a conversation; Input the sequence of keyword groups into the preset conversion model to obtain a sequence of word vector groups; Compare sequences of word vector groups of different classes, calculate similarities, and cluster the classes based on the similarities; Read the student score matrix of similar classes and make internal corrections to the student score matrix of similar classes.

4. The teaching level evaluation method based on big data according to claim 1 is characterized in that: The step of receiving the scoring information input by the students, and determining the total evaluation score by counting the scoring information and the revised student score matrix comprises: Receive scoring information input by students; The scoring information and the modified student score matrix are input into a trained evaluation model to obtain a total evaluation score; wherein the evaluation model is a preset mathematical statistical function.

5. The teaching level evaluation method based on big data according to claim 1 is characterized in that: The method further comprises: For the student score matrix of any class, calculate the mean and standard deviation of each row of data to obtain the mean column and standard deviation column; Compare the mean columns of different classes and the standard deviation columns of different classes, calculate the similarity of different classes, and cluster the classes based on the similarity; For any type of classroom, read the conversation information of each classroom, compare the conversation information, extract the majority value conversation information as the standard conversation information of this type of classroom; The majority-valued dialogue information is used to characterize the majority-valued features of dialogue information in the same class.

6. The teaching level evaluation method based on big data according to claim 5 is characterized in that: The method further comprises: Receive the acquired conversation information uploaded by a certain class; Regularly compare and match the acquired conversation information with the existing conversation information to obtain matching classes; Query the student score matrix of the matching classroom and select the optimal student score matrix; The standard dialogue information corresponding to the optimal student score matrix is ​​read as reference data and fed back to the classroom that uploaded the dialogue information.

7. A teaching level evaluation system based on big data, characterized in that: The system comprises: A dialogue information extraction module is used to record classroom interaction information based on a smart device, identify the classroom interaction information, and extract dialogue information; wherein the classroom interaction information at least includes audio information; the classroom interaction information and the extracted dialogue information have the same time period label; A matrix generation module is used to obtain a student score array containing time nodes, arrange them in chronological order, and obtain a student score matrix; the rows of the student score matrix correspond to time nodes, the columns correspond to students, and the element values ​​in the matrix represent scores; The matrix correction module is used to compare the conversation information of different classes, cluster the classes, read the student score matrix of the same class, and make internal corrections to the student score matrix of the same class; wherein the internal correction method is to reduce the difference; The evaluation total score generation module is used to receive the scoring information input by the students, and to determine the evaluation total score by counting the scoring information and the revised student score matrix.

8. The teaching level evaluation system based on big data according to claim 7 is characterized in that: The dialogue information extraction module comprises: An information recording unit, used to record classroom interaction information in real time based on smart devices; An audio feature extraction unit, used for identifying the audio features of the teacher according to preset keywords when the classroom interaction information is audio; An audio segmentation unit is used to segment the audio according to the audio features to obtain the teaching audio and the non-teaching audio; wherein the teaching audio and the non-teaching audio both contain a timestamp; The audio recognition unit is used to perform audio recognition on the audio of the teacher and the audio of the non-teacher to obtain dialogue information.

9. The teaching level evaluation system based on big data according to claim 7 is characterized in that: The matrix correction module comprises: The information conversion unit is used to perform word recognition on the dialogue information of any class, and simplify the dialogue information into a sequence of key words according to the word recognition result; each key word group corresponds to a dialogue; A word conversion unit, used to input the sequence of keyword groups into a preset conversion model to obtain a sequence of word vector groups; A comparison and clustering unit, used to compare sequences of word vector groups of different classes, calculate similarities, and cluster the classes based on the similarities; The correction execution unit is used to read the student score matrix of the same class and perform internal correction on the student score matrix of the same class.

10. The teaching level evaluation system based on big data according to claim 7 is characterized in that: The evaluation total score generating module comprises: A scoring information receiving unit, used to receive scoring information input by students; The evaluation execution unit is used to input the scoring information and the modified student score matrix into the trained evaluation model to obtain the total evaluation score; wherein the evaluation model is a preset mathematical statistical function.

Images