Teaching quality analysis system and method based on big data

Abstract

The invention discloses a teaching quality analysis system and method based on big data, and belongs to the technical field of data analys.The system calculates courseware quality scores and courseware knowledge point overdue scores through a first calculation module according to courseware knowledge points and teaching outline data in the pre-class stage; in the in-class stage, a second calculation module calculates the score of a classroom teacher according to the average volume and the average speech speed of teacher audio analyzed by an audio analysis module, and the sight drop point of each student is analyzed in real time through a sight drop point estimation module, so that the concentration state is judged. A third calculation module is used for calculating the average performance score of the students according to the concentration duration, the homework of the concentrated students in the class is screened out in the after-class stage, the comprehensive score of the student homework is calculated according to the homework score, and finally the teaching quality score is calculated through a comprehensive calculation module. According to the invention, teaching quality analysis is carried out by combining the scores before class, during class and after class, so that the accuracy and reference of the result are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a teaching quality analysis system and method based on big data. Background Art

[0002] As people pay more and more attention to education, the evaluation of teaching quality becomes particularly important. In the traditional teaching quality analysis method, questionnaires are usually used, and students evaluate and score teachers. However, the results of this method depend on the students' subjective opinions and moods, so the reference value of the analysis results is extremely low.

[0003] In the invention patent with application number CN201910165231.0, a teaching quality evaluation method is disclosed. According to at least one of the images and voices of the target teaching process, it is detected whether n evaluation conditions are met in the target teaching process, and then the teaching quality is evaluated.

[0004] In the prior art, usually only classroom teaching behavior or after-class homework test results are considered for analysis, and the after-class homework test results are greatly affected by the spontaneous concentration of classroom students and cannot directly reflect the teaching quality. Summary of the invention

[0005] In order to solve the technical problems existing in the prior art, the present invention provides a teaching quality analysis system based on big data, comprising: A database for storing student information and syllabus data; the student information includes a first face image and a corresponding student number; Courseware collection module, used to obtain the courseware of the current class; A knowledge point extraction module, used for extracting courseware knowledge points of the courseware; A first calculation module, used for calculating the courseware quality score A1 and the courseware knowledge point overdue score A2 of the courseware according to the courseware knowledge points and the syllabus data; Audio acquisition module, used to obtain the teacher's audio signal in the class and perform preprocessing; An audio analysis module, used to analyze the average volume L and average speech speed V of the preprocessed audio signal; A second calculation module is used to calculate the classroom teacher score B according to the average volume L and the average speaking speed V; A face recognition module is used to obtain the second face image of each student present in the class in real time, and to pair it with the first face image to analyze the student number of each student present; A sight point estimation module, used for analyzing the sight point of each student according to the second face image; A concentration analysis module, which is used to analyze whether each actual student is in a concentrated state according to the line-of-sight landing point and count the concentration duration ts; A third calculation module, which is used to calculate the average performance score C of students according to the concentration duration of each actual student; A screening module, which is used to screen the number r1 of concentrated students who have submitted their homework and the number r2 of qualified concentrated students according to the student numbers and the concentration duration ts of each actual student; A fourth calculation module, which is used to calculate the comprehensive score D of the homework of concentrated students; A comprehensive calculation module, which is used to calculate the teaching quality score E, and E = A1 - A2 + B + C + D.

[0006] Further, the courseware quality score A1, A1 = q1 / Q * P1; the overdue score A2 of the courseware knowledge points, A2 = q2 * K1; Wherein, P1 is the maximum score that the courseware can obtain, Q is the number of knowledge points required to be explained in the current course, q1 is the number of knowledge points required to be explained in the current course contained in the courseware, K1 is the courseware deduction coefficient, and q2 is the number of knowledge points that have exceeded the teaching deadline of the corresponding knowledge points contained in the courseware.

[0007] Further, the classroom teacher score B is specifically: B = B1 + B2; When L1 ≤ L ≤ L2, B1 = P2; when L < L1, B1 = P2 - (L1 - L) * K2; when L > L2, B1 = P2 - (L - L2) * K2; When V1 ≤ V ≤ V2, B2 = P3; when V < V1, B2 = P3 - (V1 - V) * K3; when V > V2, B2 = P3 - (V - V2) * K3; Wherein, B1 is the volume score, B2 is the speech rate score, L1 is the minimum reasonable volume threshold, L2 is the maximum reasonable volume threshold, P2 is the maximum score that the volume can obtain, K2 is the volume deduction coefficient, V1 is the minimum reasonable speech rate threshold, V2 is the maximum reasonable speech rate threshold, P3 is the maximum score that the speech rate can obtain, and K3 is the speech rate deduction coefficient.

[0008] Further, the audio analysis module includes: A feature extraction module, which is used to extract the short-time energy E1 of each frame of the preprocessed audio signal, ; A frame extraction module, which is used to extract the speech frames and silent frames of the preprocessed audio signal; A volume analysis module, which is used to calculate the average volume L according to the short-time energy of each speech frame, L = ; Wherein, ; is the short-time energy of the nth frame, N is the total number of sampling points per frame, is the amplitude of the ith sampling point in the nth frame, M is the number of frames containing speech frames in the preprocessed audio signal, is the normalized value of the short-time energy of the m-th speech frame, is the short-time energy of the m-th speech frame.

[0009] Furthermore, the audio analysis module also includes: A speech recognition module, used to convert the preprocessed audio signal into recognized text; A word segmentation technology module is used to segment the recognized text into multiple words and Chinese characters, and calculate the total word number H, which is the sum of the number of words and the number of Chinese characters; The speech rate analysis module is used to calculate the average speech rate V, V=H / T; where T is the total duration of the speech frames of the preprocessed audio signal.

[0010] Furthermore, the sight point estimation module includes: The position coordinate construction module is used to construct a classroom space coordinate system; the classroom space coordinate system takes the lower left corner of the classroom as the origin, the direction perpendicular to the podium area as the positive direction of the x-axis, the right direction as the positive direction of the y-axis, and the direction perpendicular to the ceiling as the positive direction of the z-axis.

[0011] A facial angle recognition module is used to recognize the horizontal deflection angle of each student's face in real time through the second facial image. and the vertical deflection angle of the face ; The landing point calculation module is used to analyze the y-axis sight landing point value ye and the z-axis sight landing point value ze of each student, specifically: When the student's face turns to the left, ; When the student's face turns to the right, ; When the student's face is tilted downward, ; When the student's face is tilted upward, ; in, is the y-axis sight point value of the student at the jth position, is the y-axis coordinate of the j-th position, is the lateral deflection angle of the face of the student at the jth position, is the x-axis coordinate of the plane where the podium area is located, is the x-axis coordinate of the j-th position, is the z-axis sight point value of the student at the jth position, is the z-axis coordinate of the j-th position, is the vertical deflection angle of the face of the student at the jth position.

[0012] Furthermore, the focus analysis module analyzes whether each student is in a focused state according to the sight point and calculates the focus time ts, specifically: When ymin≤ye≤ymax and zmin≤ze≤zmax, the students at the corresponding positions are in a focused state; Among them, ymax is the maximum y-axis coordinate value of the podium area, ymin is the minimum y-axis coordinate value of the podium area, zmax is the maximum z-axis coordinate value of the podium area, and zmin is the minimum z-axis coordinate value of the podium area.

[0013] Furthermore, the student average performance score C, ; Among them, R is the actual number of students, is the concentration duration of the student at the jth position, P4 is the maximum score that the student can get in class, and T is the total duration of the speech frames of the preprocessed audio signal.

[0014] Furthermore, the dedicated students are specifically the actual students corresponding to the time ratio threshold value whose ratio of ts to T is greater than or equal to the time ratio threshold value; The qualified number of focused students r2 is specifically the number of focused students who have submitted their homework and obtained a qualified score for the homework; The comprehensive score D of the student's homework is specifically: When the classroom teacher score B is greater than or equal to the teacher score threshold Bc, D=r2 / r1*P5; When the classroom teacher score B is less than the teacher score threshold Bc, ; Among them, P5 is the maximum score that can be obtained for the comprehensive score of the student's homework, and T is the total duration of the speech frames of the preprocessed audio signal.

[0015] The present invention also provides a teaching quality analysis method based on big data, comprising the following steps: S1. Obtain the courseware of the current class and extract the knowledge points of the courseware; S2, calculating the courseware quality score A1 and the courseware knowledge point overdue score A2 of the courseware; S3, focus on analyzing the students who attended the class, obtain the audio signal of the teacher in the class, pre-process it, and calculate the average volume L and average speaking speed V; S4, calculate the classroom teacher score B and the student average performance score C; S5, screening the number of focused students r1 who have submitted their homework and the number of qualified focused students r2, and calculating the comprehensive score D of the focused students' homework; S6. Calculate the teaching quality score E, E=A1-A2+B+C+D; The above-mentioned focused analysis of the actual students is as follows: S301, obtaining in real time the second face image of each student present in the class, and performing pairing analysis with the first face image to obtain the student ID of each student present; S302, analyzing the sight points of each student; S303, analyzing whether each student is in a focused state and calculating the focused time ts.

[0016] Compared with the prior art, the present invention has the following beneficial effects: Comprehensively consider the courseware quality score before class and the overdue score of courseware knowledge points, the classroom teacher score and the average student performance score during class, and the comprehensive score of focused student homework after class, so as to avoid the influence of single parameters and greatly improve the accuracy and reference of teaching quality analysis results; Logarithmically normalize the short-time energy of speech frames to eliminate amplitude differences, improve the accuracy of average volume calculation, and better match the nonlinear response of the human perception system. At the same time, it effectively compresses wide dynamic range data into a smaller range to avoid numerical overflow problems. Directly by establishing the classroom space coordinate system, according to the lateral deflection angle of each student's face and the vertical deflection angle of the face Analyze the gaze point, then analyze the concentration state and concentration duration, and then judge whether the student is a focused student by the ratio of the concentration duration to the total voice frame duration. This effectively avoids judging the student as not focusing when the teacher mutees the microphone or the student lowers his head to do in-class exercises, and further improves the accuracy of the analysis results. At the same time, the homework scores of the focused students who have submitted their homework are extracted for calculation to avoid the impact of the homework scores of non-focused students and students who were absent from class, further improving the accuracy of the analysis results. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a structural block diagram of a teaching quality analysis system based on big data of the present invention; Figure 2 It is a structural block diagram of an audio analysis module of a teaching quality analysis system based on big data of the present invention; Figure 3 It is a structural block diagram of a sight point estimation module of a teaching quality analysis system based on big data of the present invention; Figure 4 It is a flow chart of a teaching quality analysis method based on big data of the present invention; Figure 5It is a specific flow chart of the teaching quality analysis method based on big data of the present invention for conducting focused analysis on actual students. DETAILED DESCRIPTION

[0018] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Preferred embodiments of the present invention are provided in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present invention more thorough and comprehensive.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0020] See also Figure 1 As shown, the present invention provides a teaching quality analysis system based on big data, including: a database, a courseware acquisition module, a first calculation module, an audio acquisition module, an audio analysis module, a second calculation module, a face recognition module, a sight point estimation module, a focus analysis module, a third calculation module, a screening module, a fourth calculation module and a comprehensive calculation module, and a central processing unit communicatively connected to the above modules.

[0021] Database, used to store student information and syllabus data; The student information includes the first face image and the corresponding student number, name and class schedule information; The teaching syllabus data includes the knowledge points to be taught, the teaching order of the knowledge points and the teaching period corresponding to each knowledge point; each knowledge point contains one or more keywords; For example, a subject has 40 classes in a semester, and a certain knowledge point needs to be explained in the second class at the latest; The knowledge points are numbered from early to late according to the order in which the knowledge points are taught; for example, in a certain subject in a certain semester, the first knowledge point to be taught is numbered 1.

[0022] Courseware collection module, used to obtain the courseware of the current class; A knowledge point extraction module, used for extracting courseware knowledge points of the courseware; The knowledge point extraction module includes a text preprocessing module, a keyword matching module and a knowledge point matching module.

[0023] A text preprocessing module, which is used to extract the text content of the courseware and perform preprocessing to obtain a vocabulary set; the preprocessing includes word segmentation and stop word removal, etc.

[0024] Stop words refer to those words that appear very frequently but usually do not carry important information semantically. These words include common prepositions, articles, conjunctions such as "de", "he", "shi", "zai" and some pronouns. The main purposes of stop word removal are: Reduce noise: The existence of stop words often makes the text data large and redundant. Removing them can improve the processing efficiency of the algorithm; Focus on the core content: By removing stop words, one can focus more on important words and important phrases, thus better capturing the theme and meaning of the text.

[0025] Save storage space: For large-scale text databases, removing stop words helps reduce storage requirements.

[0026] The process of stop word removal is carried out according to a predefined stop word list, which varies according to different languages and their actual application scenarios. For example, in English, "the", "a", "an", "and", etc. are typical stop words; while in Chinese, "de", "le", "shi", "zai", etc. belong to this category.

[0027] A keyword matching module, which is used to calculate the cosine similarity between the words in the vocabulary set and each keyword. If the cosine similarity between a certain word and a certain keyword is greater than or equal to the first similarity threshold, then the word matches the keyword. Specifically: Convert each word and keyword into a fixed-length word vector through a pre-trained word embedding model, and then calculate the cosine value between the word and each keyword according to the word vectors of each word and keyword.

[0028] A knowledge point matching module, which is used to identify the courseware knowledge points of the courseware according to the keywords matched by the words. Specifically: According to the keywords matched by the words, query the corresponding knowledge points from the database, which are the courseware knowledge points of the courseware.

[0029] A first calculation module, which is used to calculate the courseware quality score A1 and the courseware knowledge point overdue score A2 of the courseware according to the courseware knowledge points and the teaching syllabus data; The courseware quality score A1, the specific formula is: A1 = q1 / Q * P1; Among them, P1 represents the maximum score that the courseware can obtain, which is set according to requirements; Q represents the number of knowledge points that need to be explained in the current course; q1 represents the number of knowledge points that need to be explained in the current course contained in the courseware. The knowledge points that need to be explained in the current course are obtained from the teaching syllabus data.

[0030] The overdue score of the courseware knowledge point is A2, and the specific formula is: A2=q2*K1; Among them, K1 is the courseware deduction coefficient, which can be set according to needs; q2 represents the number of knowledge points contained in the courseware that have exceeded the teaching period of the corresponding knowledge points.

[0031] Audio acquisition module, used to obtain the teacher's audio signal in the class and perform preprocessing; The audio acquisition module can select a microphone with high sensitivity and low noise coefficient to capture high-quality audio signals when the teacher is teaching.

[0032] The audio signal is preprocessed, including background noise removal, framing, windowing and other operations.

[0033] Framing is to divide a continuous audio data into short segments (i.e. "frames") so that the signal is assumed to be relatively stable in each segment. The specific steps are: Set the frame length, frame shift and sampling rate. Starting from the original audio signal, according to the set frame shift step, intercept the data segment with a length equal to the frame length; the frame length defines how many sampling points each frame contains, that is, the number of sampling points in each frame is the product of the frame length and the sampling rate; the frame shift defines the overlapping part between two adjacent frames, in order to maintain the continuity of the analysis results.

[0034] Windowing: Apply a window function to each frame of audio signal, such as Hamming window, Hann window, Blackman window, Kaiser window, etc. The purpose is to reduce the discontinuity effect and spectrum leakage problems caused by frame boundaries.

[0035] An audio analysis module, used to analyze the average volume and average speech speed of the preprocessed audio signal; See also Figure 2 As shown, the audio analysis module includes a feature extraction module, a frame extraction module, a volume analysis module, a speech recognition module, a word segmentation and counting module and a speech rate analysis module; The feature extraction module is used to extract the short-time energy E1 of each frame of the preprocessed audio signal.

[0036] Short-time energy can reflect the average strength of an audio signal in a very short period of time. The energy of an audio signal at different time points may vary greatly. By calculating the energy of each frame, the instantaneous change of the volume can be accurately captured. The specific formula is: ; in, It is represented as the short-time energy of the nth frame, N is the total number of sampling points per frame, Represented as the amplitude of the i-th sampling point in the n-th frame.

[0037] A frame extraction module, used to extract speech frames and silence frames from the preprocessed audio signal; The speech frame is a frame corresponding to an E1 greater than or equal to a first energy threshold; the speech frame indicates that the frame contains speech content; the silence frame is a frame corresponding to an E1 less than a first energy threshold; the silence frame indicates that the frame has no speech content; the first energy threshold is set according to demand.

[0038] The volume analysis module is used to calculate the average volume L based on the short-time energy of each speech frame. The specific formula is: L= ; Where M represents the number of speech frames in the preprocessed audio signal. It is represented by the normalized value of the short-time energy of the m-th speech frame; Normalizing the short-time energy can eliminate amplitude differences. The amplitude differences caused by factors such as the distance from the microphone and the speaker's volume are large. The audio clips with louder volume will dominate, affecting the accurate analysis of the average volume of the entire audio. Through normalization, it can be ensured that the energy values ​​of all frames are compared on the same scale. This scheme uses logarithmic normalization to normalize short-time energy. The specific formula is: ; It is expressed as the short-time energy of the m-th speech frame; Compared with other normalization methods, this scheme adopts logarithmic normalization, which can adapt to the perception characteristics of the human ear. The human ear's perception of sound loudness is nonlinear. By performing appropriate nonlinear transformation on short-time energy, this perception characteristic can be simulated, so that the calculated average volume is more in line with the actual human hearing perception. Therefore, logarithmic normalization can better match the nonlinear response of the human perception system; at the same time, logarithmic transformation can effectively compress wide dynamic range data into a smaller range to avoid numerical overflow problems.

[0039] A speech recognition module, used to convert the preprocessed audio signal into recognized text; The speech recognition module may select technologies such as Alibaba Cloud's intelligent speech service, Google's Speech-to-Text API, or IBM's Watson Speech to Text.

[0040] The word segmentation technology module is used to segment the recognized text into multiple words and Chinese characters, and calculate the total number of words H, where the total number of words H is the sum of the number of words and the number of Chinese characters.

[0041] The speech rate analysis module is used to calculate the average speech rate V, and the specific formula is: V = H / T; Where T is the total duration of the speech frames of the preprocessed audio signal.

[0042] The second calculation module is used to calculate the classroom teacher score B according to the average volume and the average speech rate, and the specific formula is: B = B1 + B2; Where B1 represents the volume score; B2 represents the speech rate score.

[0043] The calculation of the volume score B1 is specifically as follows: When L1 ≤ L ≤ L2, B1 = P2; when L < L1, B1 = P2 - (L1 - L) * K2; when L > L2, B1 = P2 - (L - L2) * K2; Where L1 is the minimum reasonable volume threshold, which is set according to requirements; L2 is the maximum reasonable volume threshold, which is set according to requirements; P2 is the maximum score that can be obtained for volume, which is set according to requirements; K2 is the volume deduction coefficient, which is set according to requirements.

[0044] The calculation of the speech rate score B2 is specifically as follows: When V1 ≤ V ≤ V2, B2 = P3; when V < V1, B2 = P3 - (V1 - V) * K3; when V > V2, B2 = P3 - (V - V2) * K3; Where V1 is the minimum reasonable speech rate threshold, which is set according to requirements; V2 is the maximum reasonable speech rate threshold, which is set according to requirements; P3 is the maximum score that can be obtained for speech rate, which is set according to requirements; K3 is the speech rate deduction coefficient, which is set according to requirements.

[0045] The face recognition module is used to obtain the second face images of each actually present student in the current class in real time, and perform paired analysis with the first face image to analyze the student numbers of each actually present student; The line-of-sight landing point estimation module is used to analyze the line-of-sight landing points of each actually present student according to the second face image; Refer to Figure 3 As shown, the line-of-sight landing point estimation module includes: a position coordinate construction module, a face angle recognition module, and a landing point calculation module.

[0046] The position coordinate construction module is used to construct a classroom space coordinate system. The classroom space coordinate system takes the lower left corner of the classroom as the origin, the direction perpendicular to the podium area as the positive x-axis direction, the rightward direction as the positive y-axis direction, and the direction perpendicular to the ceiling as the positive z-axis direction.

[0047] A facial angle recognition module is used to recognize the horizontal deflection angle of each student's face in real time through the second facial image. and the vertical deflection angle of the face ; It reflects the left-right rotation of the student's face. It reflects whether students raise their heads or lower their heads.

[0048] The landing point calculation module is used to analyze the y-axis sight landing point value ye and the z-axis sight landing point value ze of each student, specifically: When the student's face turns to the left, ; When the student's face turns to the right, ; When the student's face is tilted downward, ; When the student's face is tilted upward, ; in, is the y-axis sight point value of the student at the jth position, is the y-axis coordinate of the j-th position, is the lateral deflection angle of the face of the student at the jth position, is the x-axis coordinate of the plane where the podium area is located, is the x-axis coordinate of the j-th position, so we can know that That is, the vertical distance from the jth position to the plane where the podium area is located; is the z-axis sight point value of the student at the jth position, is the z-axis coordinate of the j-th position, is the vertical deflection angle of the face of the student at the jth position.

[0049] The focus analysis module is used to analyze whether each student is in a focused state based on the gaze point and calculate the focus time ts, specifically: When ymin≤ye≤ymax and zmin≤ze≤zmax, the students at the corresponding positions are in a focused state; Among them, ymax is the maximum y-axis coordinate value of the podium area, ymin is the minimum y-axis coordinate value of the podium area, zmax is the maximum z-axis coordinate value of the podium area, and zmin is the minimum z-axis coordinate value of the podium area.

[0050] The third calculation module is used to calculate the average performance score C of the students according to the concentration time of each student. The specific formula is: ; Among them, R is the actual number of students, is the concentration time of the student in the jth position; P4 is the maximum score that a student can get in class, which can be set according to needs.

[0051] The screening module is used to screen the number of focused students r1 who have submitted their homework and the number of qualified focused students r2 according to the student ID and the focused time ts of each student who has attended the class; The dedicated students are specifically the actual students whose ratio of ts to T is greater than or equal to the time ratio threshold, and the time ratio threshold is set according to the needs; By identifying the student ID on the assignment, pairing it with the student ID of the dedicated student, the number of successfully matched students is obtained, which is the number r1 of dedicated students who have submitted the assignment; The qualified number of focused students r2 is specifically the number of focused students who have submitted their homework and obtained a qualified score for the homework; The homework scores of the focused students who have submitted their homework are screened out. When the homework score is greater than or equal to the set passing threshold, the homework score is a passing score, and the number of qualified people who obtain the passing score r2 is counted.

[0052] The fourth calculation module is used to calculate the comprehensive score D of the focused student's homework, specifically: When the classroom teacher score B is greater than or equal to the teacher score threshold Bc, D=r2 / r1*P5; When the classroom teacher score B is less than the teacher score threshold Bc, ; Among them, P5 is the maximum score that can be obtained by the student's comprehensive score of the assignment, which can be set according to needs; the teacher's score threshold Bc is set according to needs.

[0053] Comprehensive calculation module, used to calculate the teaching quality score E, E=A1-A2+B+C+D.

[0054] See also Figures 4 to 5 The present invention also provides a teaching quality analysis method based on big data, which is applied to the above-mentioned teaching quality system and includes the following steps: S1. Obtain the courseware of the current class and extract the knowledge points of the courseware; S2, calculating the courseware quality score A1 and the courseware knowledge point overdue score A2 of the courseware; S3, focus on analyzing the students who attended the class, obtain the audio signal of the teacher in the class, pre-process it, and calculate the average volume L and average speaking speed V; S4, calculate the classroom teacher score B and the student average performance score C; S5, screening the number of focused students r1 who have submitted their homework and the number of qualified focused students r2, and calculating the comprehensive score D of the focused students' homework; S6. Calculate the teaching quality score E, E=A1-A2+B+C+D; The above-mentioned focused analysis of the actual students is as follows: S301, obtaining in real time the second face image of each student present in the class, and performing pairing analysis with the first face image to obtain the student ID of each student present; S302, analyzing the sight points of each student; S303, analyzing whether each student is in a focused state and calculating the focused time ts.

[0055] The specific implementation of the above steps and the specific method of acquiring data correspond to the aforementioned teaching quality system of the present invention, and will not be described in detail here.

[0056] The beneficial effects of the present invention are: Comprehensively consider the courseware quality score before class and the overdue score of courseware knowledge points, the classroom teacher score and the average student performance score during class, and the comprehensive score of focused student homework after class, so as to avoid the influence of single parameters and greatly improve the accuracy and reference of teaching quality analysis results; Logarithmically normalize the short-time energy of speech frames to eliminate amplitude differences, improve the accuracy of average volume calculation, and better match the nonlinear response of the human perception system. At the same time, it effectively compresses wide dynamic range data into a smaller range to avoid numerical overflow problems. Directly by establishing the classroom space coordinate system, according to the lateral deflection angle of each student's face and the vertical deflection angle of the face Analyze the gaze point, then analyze the concentration state and concentration duration, and then judge whether the student is a focused student by the ratio of the concentration duration to the total voice frame duration. This effectively avoids judging the student as not focusing when the teacher mutees the microphone or the student lowers his head to do in-class exercises, and further improves the accuracy of the analysis results. At the same time, the homework scores of the focused students who have submitted their homework are extracted for calculation to avoid the impact of the homework scores of non-focused students and students who were absent from class, further improving the accuracy of the analysis results.

[0057] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0058] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0059] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0060] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, read-only memory), random access memory (RAM, random access memory), disk or optical disk and other media that can store program code.

[0061] In the description of the specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0062] The above contents are merely examples and explanations of the structure of the present invention. The technicians in this technical field may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the structure of the invention or exceed the scope defined by the claims, they should all fall within the protection scope of the present invention.

Claims

1. A teaching quality analysis system based on big data, characterized in that: Including: A database for storing student information and syllabus data; the student information includes a first face image and the corresponding student ID. A courseware acquisition module for obtaining the courseware of the current class. A knowledge point extraction module for extracting the knowledge points of the courseware. A first calculation module for calculating the courseware quality score A1 and the overdue score A2 of the courseware knowledge points according to the courseware knowledge points and the syllabus data. An audio acquisition module for acquiring the audio signal of the teacher in the current class and performing preprocessing. An audio analysis module for analyzing the average volume L and the average speech rate V of the preprocessed audio signal. A second calculation module for calculating the classroom teacher score B according to the average volume L and the average speech rate V. A face recognition module for obtaining the second face images of the actual students in the current class in real time and performing paired analysis with the first face image to analyze the student IDs of the actual students. A gaze landing point estimation module for analyzing the gaze landing points of the actual students according to the second face images. A concentration analysis module for analyzing whether the actual students are in a concentrated state according to the gaze landing points and counting the concentration duration ts. A third calculation module for calculating the average student performance score C according to the concentration duration of each actual student. A screening module for screening the number r1 of concentrated students who have submitted their homework and the number r2 of qualified concentrated students according to the student IDs and the concentration duration ts of each actual student. A fourth calculation module for calculating the comprehensive score D of the concentrated students' homework. A comprehensive calculation module for calculating the teaching quality score E, where E = A1 - A2 + B + C + D.

2. The teaching quality analysis system based on big data according to claim 1 is characterized in that: The courseware quality score A1, A1 = q1 / Q * P1; the overdue score A2 of the courseware knowledge points, A2 = q2 * K1. Wherein, P1 is the maximum score that the courseware can obtain, Q is the number of knowledge points required to be explained in the current course, q1 is the number of knowledge points required to be explained in the current course contained in the courseware, K1 is the courseware deduction coefficient, and q2 is the number of knowledge points that have exceeded the corresponding knowledge point teaching deadline contained in the courseware.

3. The teaching quality analysis system based on big data according to claim 1 is characterized in that: The classroom teacher score B is specifically: B = B1 + B2. When L1 ≤ L ≤ L2, B1 = P2; when L < L1, B1 = P2 - (L1 - L) * K2; when L > L2, B1 = P2 - (L - L2) * K2. When V1 ≤ V ≤ V2, B2 = P3; when V < V1, B2 = P3 - (V1 - V) * K3; when V > V2, B2 = P3 - (V - V2) * K3. Wherein, B1 is the volume score, B2 is the speech rate score, L1 is the minimum reasonable volume threshold, L2 is the maximum reasonable volume threshold, P2 is the maximum score that the volume can obtain, K2 is the volume deduction coefficient, V1 is the minimum reasonable speech rate threshold, V2 is the maximum reasonable speech rate threshold, P3 is the maximum score that the speech rate can obtain, and K3 is the speech rate deduction coefficient.

4. The teaching quality analysis system based on big data according to claim 3 is characterized in that: The audio analysis module includes: The feature extraction module is used to extract the short-time energy E1 of each frame of the preprocessed audio signal. ; A frame extraction module for extracting the speech frames and silent frames of the preprocessed audio signal. The volume analysis module is used to calculate the average volume L according to the short-time energy of each speech frame, L= ; in, ; is the short-time energy of the nth frame, N is the total number of sampling points per frame, is the amplitude of the ith sampling point in the nth frame, M is the number of frames containing speech frames in the preprocessed audio signal, is the normalized value of the short-time energy of the m-th speech frame, is the short-time energy of the m-th speech frame.

5. The teaching quality analysis system based on big data according to claim 4 is characterized in that: The audio analysis module further includes: A speech recognition module for converting the preprocessed audio signal into recognized text. A word segmentation technology module is used to segment the recognized text into multiple words and Chinese characters, and calculate the total word number H, which is the sum of the number of words and the number of Chinese characters; The speech rate analysis module is used to calculate the average speech rate V, V=H / T; where T is the total duration of the speech frames of the preprocessed audio signal.

6. The teaching quality analysis system based on big data according to claim 1 is characterized in that: The sight point estimation module comprises: A position coordinate construction module is used to construct a classroom space coordinate system; the classroom space coordinate system takes the lower left corner of the classroom as the origin, the direction perpendicular to the podium area as the positive direction of the x-axis, the right direction as the positive direction of the y-axis, and the direction perpendicular to the ceiling as the positive direction of the z-axis; A facial angle recognition module is used to recognize the horizontal deflection angle of each student's face in real time through the second facial image. and the vertical deflection angle of the face ; The landing point calculation module is used to analyze the y-axis sight landing point value ye and the z-axis sight landing point value ze of each student, specifically: When the student's face turns to the left, ; When the student's face turns to the right, ; When the student's face is tilted downward, ; When the student's face is tilted upward, ; in, is the y-axis sight point value of the student at the jth position, is the y-axis coordinate of the j-th position, is the lateral deflection angle of the face of the student at the jth position, is the x-axis coordinate of the plane where the podium area is located, is the x-axis coordinate of the j-th position, is the z-axis sight point value of the student at the jth position, is the z-axis coordinate of the j-th position, is the vertical deflection angle of the face of the student at the jth position.

7. The teaching quality analysis system based on big data according to claim 6 is characterized in that: The concentration analysis module analyzes whether each student is in a state of concentration according to the sight point and calculates the concentration time ts, specifically: When ymin≤ye≤ymax and zmin≤ze≤zmax, the students at the corresponding positions are in a focused state; Among them, ymax is the maximum y-axis coordinate value of the podium area, ymin is the minimum y-axis coordinate value of the podium area, zmax is the maximum z-axis coordinate value of the podium area, and zmin is the minimum z-axis coordinate value of the podium area.

8. The teaching quality analysis system based on big data according to claim 7 is characterized in that: The average performance score of the students is C, ; Among them, R is the actual number of students, is the concentration duration of the student at the jth position, P4 is the maximum score that the student can get in class, and T is the total duration of the speech frames of the preprocessed audio signal.

9. The teaching quality analysis system based on big data according to claim 1 is characterized in that: The dedicated students are specifically the students who attend school when the ratio of ts to T is greater than or equal to the time ratio threshold; The qualified number of focused students r2 is specifically the number of focused students who have submitted their homework and obtained a qualified score for the homework; The comprehensive score D of the student's homework is specifically: When the classroom teacher score B is greater than or equal to the teacher score threshold Bc, D=r2 / r1*P5; When the classroom teacher score B is less than the teacher score threshold Bc, ; Among them, P5 is the maximum score that can be obtained for the comprehensive score of the student's homework, and T is the total duration of the speech frames of the preprocessed audio signal.

10. A teaching quality analysis method based on big data, applied to the teaching quality analysis system according to any one of claims 1 to 9, characterized in that: The following steps are involved: S1. Obtain the courseware of the current class and extract the knowledge points of the courseware; S2, calculating the courseware quality score A1 and the courseware knowledge point overdue score A2 of the courseware; S3, focus on analyzing the students who attended the class, obtain the audio signal of the teacher in the class, pre-process it, and calculate the average volume L and average speaking speed V; S4, calculate the classroom teacher score B and the student average performance score C; S5, screening the number of focused students r1 who have submitted their homework and the number of qualified focused students r2, and calculating the comprehensive score D of the focused students' homework; S6. Calculate the teaching quality score E, E=A1-A2+B+C+D; The above-mentioned focused analysis of the actual students is as follows: S301, obtaining in real time the second face image of each student present in the class, and performing pairing analysis with the first face image to obtain the student ID of each student present; S302, analyzing the sight points of each student; S303, analyzing whether each student is in a focused state and calculating the focused time ts.

Images