Teaching assisting method

Abstract

The invention provides a teaching assistance method. The method comprises the steps of determining scores of a teacher under a plurality of classroom indexes in a first classroom based on audio information and behavior information in the first classroom; determining whether the teacher is a to-be-helped teacher or not according to the scores of the teacher under the plurality of classroom indexes; and if the teacher is the teacher to be helped, determining a recommended famous teacher corresponding to the teacher to be helped from the famous teacher library based on the score of the teacher to be helped under the plurality of classroom indexes, and recommending the recommended famous teacher to the teacher to be helped so as to perform teaching assistance on the teacher to be helped by using the recommended famous teacher, thereby realizing objective and comprehensive evaluation on the teaching level of the teacher. Therefore, accurate famous teacher recommendation is provided for teachers needing assistance, the quality and the teaching level of the whole teacher team can be improved, and powerful support is provided for improving the teaching quality and promoting teaching innovation.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to a teaching assistance method. Background Art

[0002] In the field of education, the same course with different teaching methods is widely used as a teaching model. The same course with different teaching methods means that different teachers use different teaching methods and strategies to teach the same teaching content, aiming to promote teaching innovation and improve teaching quality by comparing the teaching effects of different teachers.

[0003] At present, the relevant technologies for evaluating teachers' teaching quality mainly rely on student evaluations, peer reviews, and subjective evaluations by teaching management departments.

[0004] However, although these methods in related technologies can reflect the teaching level of teachers to a certain extent, they have the problems of strong subjectivity, inconsistent standards, lack of objectivity, etc. In addition, some schools or institutions also use methods such as video recording analysis and classroom observation, but these methods are often time-consuming and laborious, and it is difficult to fully and accurately capture the teaching characteristics and problems of teachers. Summary of the invention

[0005] The present invention provides a teaching assistance method.

[0006] According to a first aspect of the present disclosure, a teaching assistance method is provided, the method comprising: Based on the audio information and behavior information in the first classroom, determining the scores of the teacher in the first classroom under multiple classroom indicators; Determine whether a teacher is in need of assistance based on the teacher's scores on multiple classroom indicators; If the teacher is a teacher to be assisted, the recommended master teacher corresponding to the teacher to be assisted will be determined from the master teacher database based on the scores of the teacher to be assisted under multiple classroom indicators, and the recommended master teacher will be recommended to the teacher to be assisted, so that the recommended master teacher can provide teaching assistance to the teacher to be assisted.

[0007] In some embodiments of the present disclosure, the multiple classroom indicators include a classroom content quality indicator, and based on the audio information and behavior information in the first classroom, the score of each teacher in the first classroom under the multiple classroom indicators is determined, including: Determine the scores of the students' in-class tests corresponding to the first class; Obtain classroom performance scores corresponding to behavioral information; Based on students' in-class test scores and classroom performance scores, determine the teacher's classroom content quality score under the classroom content quality index.

[0008] In some embodiments of the present disclosure, determining the scores of the student in-class test corresponding to the first class includes: According to the following formula 1, the scores of the students' in-class test corresponding to the first class are determined:

[0009]

[0010] Formula 1 in, is the relative score of the class test of the i-th student, is the absolute score of the class test of the i-th student, is the mean absolute score of students’ in-class tests, is the standard deviation of the students’ absolute scores in the in-class tests, N is the number of students, Q is the mean of the students’ relative scores in the in-class tests, Scoring students' in-class tests, is the preset weight, The default basic score is: According to the following formula 2, based on the students' in-class test scores and classroom performance scores, the teacher's classroom content quality score under the classroom content quality indicator is determined: Formula 2 in, Rate the quality of the course content, Scoring students' in-class tests, The preset weights corresponding to the students' in-class test scores, Grade class performance. Preset weights for grading class performance.

[0011] In some embodiments of the present disclosure, the multiple classroom indicators include a classroom interaction indicator, and based on the audio information and behavior information in the first classroom, determining the score of each teacher in the first classroom under the multiple classroom indicators includes: Determine the learning interaction behavior and teaching interaction behavior corresponding to the behavior information; Based on learning interaction behaviors and teaching interaction behaviors, determine the teacher behavior occupancy rate and teacher-student behavior conversion rate; Based on the teacher behavior occupancy rate and the teacher-student behavior conversion rate, determine the class type corresponding to the teacher; Based on the class type, determine the teacher's classroom interaction score under the classroom interaction indicator.

[0012] In some embodiments of the present disclosure, based on the teacher behavior occupancy rate and the teacher-student behavior conversion rate, determining the teacher's classroom interaction score under the classroom interaction index includes: If the teacher behavior occupancy rate is less than or equal to the first preset threshold, the class type corresponding to the teacher is determined to be a practice type, and the teacher's classroom interaction score under the classroom interaction index is determined according to the following formula 3: Formula 3 in, Score class interactions, For teacher behavior share, is a first preset threshold; If the teacher behavior occupancy rate is greater than or equal to the second preset threshold, the class type corresponding to the teacher is determined to be a lecture type, and the classroom interaction score of the teacher under the classroom interaction index is determined according to the following formula 4: Formula 4 in, Score class interactions, For teacher behavior share, is a second preset threshold; If the teacher-student behavior conversion rate is greater than or equal to the third preset threshold, the class type corresponding to the teacher is determined to be a dialogue type. If the teacher behavior occupancy rate is greater than the first preset threshold and less than the second preset threshold, and the teacher-student behavior conversion rate is less than the third preset threshold, the class type corresponding to the teacher is determined to be a mixed type. According to the following formula 5, the classroom interaction score of the teacher under the classroom interaction index is determined:

[0013] Formula 5 in, is the classroom interaction score, E is the interactive behavior effectiveness score, S is the effective interactive behavior sequence, is the preset weight corresponding to the effective interactive behavior sequence, To preset adjustment parameters, is the frequency of occurrence of effective interactive behavior sequence S, is the duration of the effective interactive behavior sequence S, The total duration of the first class.

[0014] In some embodiments of the present disclosure, the classroom indicator includes a classroom fluency indicator, and based on the audio information and behavior information in the first classroom, determining the score of each teacher in the first classroom under the multiple classroom indicators includes: Obtaining an audio feature vector corresponding to the audio information, and performing mapping processing on the audio feature vector to obtain a disfluency label corresponding to the audio feature vector; For the non-smooth events corresponding to the non-smooth labels of each frame, determine the occurrence frequency of each non-smooth event; Based on the frequency of each disfluency event, determine the negative impact score of each disfluency event on class fluency; Based on the negative impact score, determine the teacher's classroom fluency score under the classroom fluency index.

[0015] In some embodiments of the present disclosure, based on the occurrence frequency of each non-fluency event, determining the negative impact score of each non-fluency event on class fluency includes: According to the following formula 6, the negative impact score of each disfluency event on class fluency is determined based on the frequency of occurrence of each disfluency event: Formula 6 in, is the negative impact score of the disfluency event Ck on the fluency of the class, Ck is the disfluency event corresponding to the disfluency label k, is the preset weight corresponding to the non-smooth label k, To preset adjustment parameters, is the occurrence frequency of the non-smooth event Ck, is the total number of windows corresponding to the audio information; Based on the negative impact score, the teacher's class fluency score under the class fluency index is determined according to the following formula 7: Formula 7 in, Score the fluency of the class, T is the total score of the preset threshold, Score negative impact.

[0016] In some embodiments of the present disclosure, determining whether a teacher is a teacher to be assisted is performed based on the teacher's scores under multiple classroom indicators, including: The teacher's scores under different classroom indicators are counted to obtain the teacher's comprehensive score; If the comprehensive score and the scoring results corresponding to each classroom indicator under different classroom indicators do not meet the expected conditions, the teacher will be determined as a teacher to be assisted. The expected conditions are that the comprehensive score is greater than or equal to the comprehensive threshold, and the scoring results corresponding to each classroom indicator under different classroom indicators are greater than or equal to the scoring threshold corresponding to each classroom indicator.

[0017] In some embodiments of the present disclosure, if a teacher is a teacher to be assisted, a recommended famous teacher corresponding to the teacher to be assisted is determined from a famous teacher database based on the scores of the teacher to be assisted under multiple classroom indicators, including: Determine the classroom support indicators and support intensity corresponding to the teachers to be supported; Using the support intensity, support classroom indicators and the preset scoring system, determine the support expectations of each famous teacher in the famous teacher database. Based on the expected support score and the effective range of the preset support learning curve, the famous teachers in the famous teacher database are matched and filtered to obtain multiple pre-recommended famous teachers; Determine the compatibility between the teacher to be assisted and multiple pre-recommended famous teachers, and based on the compatibility, select the recommended famous teacher corresponding to the teacher from multiple pre-recommended famous teachers.

[0018] In some embodiments of the present disclosure, the support intensity, support classroom indicators and a preset scoring system are used to determine the support expectation score of each famous teacher in the famous teacher database for the teacher, including: According to the following formula 8, using the support intensity, support classroom indicators and the preset scoring system, the expected support score of each famous teacher in the famous teacher database for teachers is determined: Formula 8 in, is the score of famous teacher M under the supporting classroom indicator i, is the score of teacher N under the support classroom indicator i; is the score of the famous teacher M's expectation of helping teachers, k is the preset factor, is the support weight of supporting classroom indicator i.

[0019] The teaching assistance method provided by the present invention determines the scores of the teacher in the first classroom under multiple classroom indicators based on the audio information and behavior information in the first classroom; determines whether the teacher is a teacher to be assisted based on the scores of the teacher under multiple classroom indicators; if the teacher is a teacher to be assisted, then based on the scores of the teacher to be assisted under multiple classroom indicators, determines the recommended famous teacher corresponding to the teacher to be assisted from the famous teacher database, and recommends the recommended famous teacher to the teacher to be assisted, so as to use the recommended famous teacher to provide teaching assistance to the teacher to be assisted, realize an objective and comprehensive evaluation of the teacher's teaching level, and provide accurate famous teacher recommendations for teachers in need of assistance based on this, which is helpful to improve the quality and teaching level of the overall teaching team, and provide strong support for improving teaching quality and promoting teaching innovation.

[0020] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure. Figure 1 A flow chart of a teaching assistance method provided by an embodiment of the present disclosure; Figure 2 A flowchart of another teaching assistance method provided by an embodiment of the present disclosure; Figure 3A flowchart of another teaching assistance method provided by an embodiment of the present disclosure; Figure 4 A flowchart of another teaching assistance method provided by an embodiment of the present disclosure; Figure 5 An event distribution diagram of a specific non-smooth event provided by an embodiment of the present disclosure; Figure 6 A flowchart of another teaching assistance method provided by an embodiment of the present disclosure; Figure 7 A flowchart of another teaching assistance method provided by an embodiment of the present disclosure; Figure 8 A schematic diagram of the cumulative difference distribution of teachers to be assisted provided by an embodiment of the present disclosure; Fig. 9 A schematic diagram of a learning curve change assistance provided by an embodiment of the present disclosure Fig.10 A structural schematic diagram of a teaching assistance device provided in an embodiment of the present disclosure; Fig.11 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0022] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0023] Same course, different structure is widely used as an innovative teaching model. It requires different teachers to design and implement different teaching methods and strategies for the same teaching content, with the aim of promoting teaching innovation by comparing teaching effects, thereby improving the overall teaching quality.

[0024] At present, the relevant technologies for evaluating the teaching quality of teachers mainly rely on student evaluation, peer review and subjective evaluation of teaching management departments. Although these methods can reflect the teaching level of teachers to a certain extent, they have obvious limitations such as strong subjectivity, inconsistent standards and lack of objectivity. This may lead to inaccurate evaluation results and make it difficult to fully reflect the real teaching ability of teachers.

[0025] In order to make up for the shortcomings of the above evaluation methods, some schools or institutions also use more objective means such as video recording analysis and classroom observation. However, these methods are often time-consuming and laborious in implementation, and it is difficult to fully and accurately capture the teaching characteristics and potential problems of teachers.

[0026] In order to solve the problems in related technologies, the teaching assistance method proposed in the present invention evaluates the teacher's performance on multiple classroom indicators by comprehensively analyzing the audio and behavioral information in the first classroom, and then determines whether assistance is needed, and accurately recommends suitable famous teachers from the famous teacher database for assistance, so as to objectively and comprehensively improve the teacher's teaching level and promote teaching quality and teaching innovation.

[0027] The following describes the teaching assistance method, device, electronic device, storage medium and computer program product of the embodiments of the present disclosure with reference to the accompanying drawings.

[0028] Figure 1 The following is a flow chart of a teaching assistance method provided by an embodiment of the present disclosure. Figure 1 As shown, the method includes: Step 101: Based on the audio information and behavior information in the first classroom, determine the scores of the teacher in the first classroom under multiple classroom indicators.

[0029] In some embodiments, the audio information may include the teacher's teaching audio text information and the student's learning audio text information; the behavior information may include the teacher's teaching behavior information and the student's learning behavior information.

[0030] In order to comprehensively and accurately evaluate the teacher's teaching performance in the first classroom, the present disclosure can collect audio data and video data from the first classroom, the audio data including the teacher's teaching audio data and the student's learning audio data; perform speech recognition on the teaching audio data and the learning audio data respectively to obtain the teacher's teaching audio text information and the student's learning audio text information); extract video images from the video data at preset time intervals; use a deep learning model to perform behavior detection on the teacher and the student in the video image respectively to obtain the teacher's teaching behavior characteristics and the student's learning behavior characteristics; use a deep learning model, combined with the teaching audio text information and the learning audio text information, to determine the teacher's teaching behavior information and the student's learning behavior information.

[0031] Among them, the audio data includes sound records of teachers’ teaching explanations and students’ learning discussions, providing rich voice materials; the video data captures the real-time images in the first classroom and is an important basis for analyzing teacher-student behavior.

[0032] Specifically, the present disclosure can use the all-in-one Windows client and an external array microphone to record audio and video through the ffmpeg tool. During the recording process, the video can be divided into ts file segments and stored, and these segments are uploaded to OBS (object storage service) in near real time. After the recording is completed, the m3u8 index file is automatically generated and the backend is notified for subsequent processing.

[0033] In terms of audio processing, the present disclosure can first extract the audio track from the recorded audio and video files, and use tools such as FFmpeg to generate independent WAV format audio files, i.e., audio data. Subsequently, the audio data is preprocessed, including noise reduction, removal of silent segments, equalization of volume, etc., to improve the audio quality. Then, the key hot words in the classroom are extracted using the lesson plan recognition results as auxiliary inputs for the speech recognition model. The audio data is subjected to speech recognition through the Wav2Vec and Comformer models to generate text content, and the speaker is perceived using the Wav2Vec and Transformer models. Finally, structured text data containing information such as speakers, timestamps, content, etc. is generated, i.e., the audio information in the present disclosure.

[0034] In terms of video processing, the present disclosure can extract a frame from the video at every preset time interval (for example, 5 seconds) to generate a series of static video images. These video images are preprocessed, including image enhancement, denoising, scaling, etc., to improve image quality. A deep learning model is used to detect and analyze the teaching behavior characteristics of teachers (such as teachers' body language, such as gestures, standing posture, movement, etc.) and the learning behavior characteristics of students (such as students' expressions and attention status, etc.). Combining the teaching behavior characteristics and learning behavior characteristics of the video image with the audio information, a deep learning model is used to determine the behavior information (i.e., the teaching behavior information and learning behavior information of teachers and students, such as teacher questions, teacher instructions, teacher summaries, teacher inspections and guidance, etc.).

[0035] Specifically, the behavior information in the first classroom can be determined based on the mapping relationship between the behavior characteristics (teaching behavior characteristics and learning behavior characteristics) and the behavior information (teaching behavior information and learning behavior information) in Table 1.

[0036]

[0037] Table 1 In the evaluation process, the disclosure uses multiple classroom indicators to comprehensively measure the teaching quality of teachers. The multiple classroom indicators in the disclosure may include classroom content quality indicators, classroom interaction indicators, and classroom fluency indicators, which can comprehensively reflect the teaching ability and classroom performance of teachers.

[0038] For the above-mentioned multiple classroom indicators, the present disclosure can specifically determine the score under each classroom indicator by the following method, namely, determining the total score of the students' in-class tests, and determining the teacher's classroom content quality score under the classroom content quality indicator based on the total score of the in-class tests and the behavior information; using the behavior analysis method to determine the teacher behavior occupancy rate and the teacher-student behavior conversion rate corresponding to the behavior information, and using the teacher behavior occupancy rate and the teacher-student behavior conversion rate to determine the teacher's classroom interaction score under the classroom interaction indicator; determining the non-fluid events corresponding to the audio information, and determining the teacher's classroom fluency score under the classroom fluency indicator based on the non-fluid events.

[0039] Step 102: Determine whether the teacher is a teacher who needs assistance based on the teacher's scores under multiple classroom indicators.

[0040] In some embodiments, the present disclosure can obtain a comprehensive score of the teacher after obtaining the scores of the teacher under the three classroom indicators of course content quality, classroom interaction, and classroom fluency. Based on the comprehensive score and the score under each classroom indicator, combined with the preset comprehensive threshold and score threshold, it is determined whether the current teacher is a teacher to be assisted. The teacher to be assisted is a teacher who needs teaching assistance.

[0041] The comprehensive threshold and the scoring threshold can be set according to the actual situation of the school or educational institution, and are not limited in the embodiments of the present disclosure. The present disclosure compares the comprehensive score with the scoring pre-comprehensive threshold and the scoring threshold under each classroom indicator, and determines whether the teacher is listed as a target for assistance based on the comparison result.

[0042] Step 103: If the teacher is a teacher to be assisted, based on the scores of the teacher to be assisted under multiple classroom indicators, a recommended master teacher corresponding to the teacher to be assisted is determined from a database of master teachers, and the recommended master teacher is recommended to the teacher to be assisted, so that the recommended master teacher can provide teaching assistance to the teacher to be assisted.

[0043] In some embodiments, after a teacher is determined to be a teacher in need of assistance, the present disclosure can screen out the most suitable recommended teacher from the famous teacher database based on the teacher's scores under multiple classroom indicators, and recommend this famous teacher to the teacher in need of assistance in order to provide targeted teaching assistance.

[0044] Specifically, the present disclosure can identify which classroom indicators the teachers to be assisted perform weakly on based on their scores on classroom indicators such as course content quality, classroom interaction, and classroom fluency, and use these weak classroom indicators as indicators for assistance classes.

[0045] According to the supporting classroom indicators of the teachers to be assisted, pre-recommended famous teachers with relevant teaching experience and excellent performance are selected from the famous teacher database. The famous teacher database may include information such as teaching expertise, success cases, and student evaluations of multiple famous teachers for accurate matching. Among the pre-recommended famous teachers selected, the present disclosure can select a pre-recommended famous teacher who best matches the needs of the teacher to be assisted as a recommended famous teacher. The selected recommended famous teacher information is provided to the teachers to be assisted, including the famous teacher's teaching background, success cases, etc., so as to formulate a detailed assistance plan for the teachers to be assisted.

[0046] Among them, during the assistance process, the present disclosure can also regularly track the progress of the teachers to be assisted and re-evaluate them. According to the evaluation results, timely recommendations and optimizations are made, and the assistance plan is adjusted to ensure the maximum assistance effect.

[0047] Through the above steps, the disclosure can ensure that all teachers to be assisted in the first classroom can obtain precise assistance from famous teachers, thereby effectively improving their teaching level and improving teaching quality.

[0048] In summary, the technical solution provided by the present invention determines the scores of the teacher in the first classroom under multiple classroom indicators based on the audio information and behavioral information in the first classroom; determines whether the teacher is a teacher to be assisted based on the scores of the teacher under multiple classroom indicators; if the teacher is a teacher to be assisted, then based on the scores of the teacher to be assisted under multiple classroom indicators, determines the recommended famous teachers corresponding to the teacher to be assisted from the famous teacher database, and recommends the recommended famous teachers to the teacher to be assisted, so as to use the recommended famous teachers to provide teaching assistance to the teachers to be assisted, achieve an objective and comprehensive evaluation of the teachers' teaching level, and provide accurate famous teacher recommendations for teachers in need of assistance based on this, which is helpful to improve the quality and teaching level of the overall teaching team, and provide strong support for improving teaching quality and promoting teaching innovation.

[0049] As a possible implementation, Figure 2 The flowchart of another teaching assistance method shown in the figure, based on the above embodiment, the multiple classroom indicators include a classroom content quality indicator, and the specific process of determining the score of each teacher in the first classroom under the multiple classroom indicators based on the audio information and behavior information in the first classroom includes the following steps: Step 201: Determine the scores of the students' in-class test corresponding to the first class.

[0050] In some embodiments, the scores of students' in-class tests can intuitively reflect the learning feedback of the current classroom knowledge points. Generally speaking, the better the data of the in-class test scores, the more it can reflect the situation of students' absorption of the course, and the better the quality of the current course content can be intuitively shown. However, from an objective perspective, there is a certain subjective influence on the data of the in-class test scores, including the teacher's test question design and the difficulty of the knowledge points. For this reason, in order to more reasonably display the scores of students' in-class tests, the present disclosure can use the relative scores of in-class tests to better measure the authenticity of the scores of in-class tests. That is, the present disclosure can determine the relative scores of students' in-class tests according to the absolute scores of students' in-class tests; and determine the scores of students' in-class tests according to the relative scores of students' in-class tests.

[0051] Specifically, the present disclosure may determine the student in-class test score corresponding to the first class according to the following formula 1: Formula 1 in, is the relative score of the class test of the i-th student, is the absolute score of the class test of the i-th student, is the mean absolute score of students’ in-class tests, is the standard deviation of the students’ absolute scores in the in-class tests, N is the number of students, Q is the mean of the students’ relative scores in the in-class tests, Scoring students' in-class tests, is the preset weight, The default basic score.

[0052] Step 202: Obtain the classroom performance score corresponding to the behavior information.

[0053] In some embodiments, the present disclosure can score the classroom behavior performance of the teacher being evaluated (i.e., the teacher in the first classroom) from multiple evaluation dimensions such as learning objectives, learning before teaching, teaching based on learning, effective interaction, and on-the-spot training, and obtain the classroom performance score corresponding to the teacher's behavior information. That is, the present disclosure can determine the evaluation dimension corresponding to the behavior information based on the teacher's behavior information, and determine the teacher's classroom performance score based on the mapping relationship between the evaluation dimension and the preset score.

[0054] Among them, the classroom performance score of the teacher in this disclosure can also be determined by the teachers who listen to the class directly based on the teacher's behavior information and the mapping relationship between the evaluation dimensions and the preset scores.

[0055] The mapping relationship between each evaluation dimension and the preset score can be shown in Table 2.

[0056]

[0057] Table 2 Step 203: Based on the students' in-class test scores and classroom performance scores, determine the teacher's classroom content quality score under the classroom content quality index.

[0058] In some embodiments, the present disclosure may conduct a comprehensive assessment of the quality of a teacher's teaching content based on students' in-class test scores and classroom performance scores, that is, determine the teacher's classroom content quality score under the classroom content quality index.

[0059] The mean and dispersion of students' in-class test scores can truly reflect the intuitive help of the course content to students. The teacher's classroom performance score can standardize the compliance of the teacher's content quality assessment.

[0060] Before determining the teacher's classroom content quality score, the present disclosure can perform data preprocessing on the in-class test scores and classroom performance scores to remove invalid or abnormal data points, which helps to reduce the impact of noise on the evaluation results. The specific data preprocessing steps are based on actual needs and are not limited in the embodiments of the present disclosure.

[0061] Specifically, the present disclosure can determine the teacher's classroom content quality score under the classroom content quality index based on the student's in-class test score and classroom performance score according to the following formula 2: Formula 2 in, Rate the quality of the course content, Scoring students' in-class tests, The preset weights corresponding to the students' in-class test scores, Grade class performance. Preset weights for grading class performance.

[0062] Among them, the size of the preset weights (i.e., the preset weights corresponding to the students' in-class test scores and the preset weights corresponding to the classroom performance scores) can be determined according to their importance when evaluating the quality of the teacher's teaching content to ensure that they can reasonably reflect the actual contribution of each indicator during the evaluation process, and are not limited in the present disclosed embodiments.

[0063] In summary, this disclosure can combine these two aspects of data to give teachers a comprehensive and accurate classroom content quality score under the classroom content quality index by accurately calculating the scores of students' in-class tests in the first class and obtaining the classroom performance scores corresponding to the teacher's behavior information in the class. The classroom content quality score not only reflects the students' mastery of the classroom content, but also reflects the teacher's teaching effect and classroom management ability, providing a strong and objective basis for the continuous improvement of teaching quality.

[0064] As a possible implementation, Figure 3 The flowchart of another teaching assistance method shown in the figure, based on the above embodiment, the multiple classroom indicators include classroom interaction indicators, and the specific process of determining the score of each teacher in the first classroom under the multiple classroom indicators based on the audio information and behavior information in the first classroom includes the following steps: Step 301: Determine the learning interaction behavior and teaching interaction behavior corresponding to the behavior information.

[0065] In some embodiments, in order to comprehensively evaluate the teacher's classroom interaction performance under the classroom interaction index, the present disclosure can filter out the interactive behavior information related to learning interaction and teaching interaction from the behavioral information based on the mapping relationship between the behavioral characteristics and the behavioral information, and clearly encode the interactive behavior information. After obtaining the interactive behavior information, the present disclosure can determine the student's learning interactive behavior and the teacher's teaching interactive behavior from the interactive behavior information.

[0066] Among them, the interactive behavior information, interactive behavior features and corresponding codes in this disclosure are specifically shown in Table 3.

[0067]

[0068] Table 3 This disclosure is to intuitively display the interaction between teachers and students in the first classroom, and specifically the ST analysis method can be used. This method can clearly divide the interactive behavior information into learning interactive behavior (S) and teacher interactive behavior (T). Among them, teaching interactive behavior T refers to the teacher's information transmission behavior, such as teacher explanation, demonstration, experiment, blackboard writing, prompting, questioning, evaluation, summary, etc. Learning interactive behavior S refers to all behaviors other than teaching interactive behavior T, such as taking notes, doing experiments, doing homework, etc.

[0069] Specifically, the present disclosure can use the ST analysis method to sample the interactive behavior information at preset time intervals to determine the learning interactive behavior and teacher interactive behavior in the interactive behavior information. It can be understood that since there can be multiple teachers teaching in the first class, in order to provide precise assistance to each teacher, the present disclosure can mark the learning interactive behavior and teacher interactive behavior in the interactive behavior information corresponding to each teacher according to the behavior subject, as shown in Tables 4 and 5. Table 4 is a learning interactive behavior S and teacher interactive behavior T of teacher A obtained by taking the first 10 minutes of the first class as an example, and Table 5 is a learning interactive behavior S and teacher interactive behavior T of teacher B obtained by taking the first 10 minutes of the first class as an example.

[0070]

[0071] Table 4

[0072] Table 5 Step 302: Based on the learning interaction behavior and the teaching interaction behavior, determine the teacher behavior occupancy rate and the teacher-student behavior conversion rate.

[0073] In some embodiments, the present disclosure may adopt ST analysis method to calculate the teacher behavior occupancy rate (Rt) and the teacher-student behavior conversion rate (Ch) based on the learning interaction behavior (S) and the teaching interaction behavior (T). That is, the present disclosure may determine the total number of the teacher's teaching interaction behaviors and the total number of interactive behaviors of the learning interaction behavior and the teaching interaction behavior based on the learning interaction behavior and the teaching interaction behavior; traverse the character string composed of the learning interaction behavior and the teaching interaction behavior to determine the total number of continuous behaviors corresponding to the character string, and the total number of continuous behaviors refers to the total number of identical or different behavior sequences that appear continuously in the character string composed of the learning interaction behavior and the teaching interaction behavior; determine the teacher behavior occupancy rate and the teacher-student behavior conversion rate based on the total number of teaching interaction behaviors, the total number of interactive behaviors and the total number of continuous behaviors.

[0074] Specifically, after obtaining the learning interaction behavior S and the teaching interaction behavior T, the present disclosure can count the total number of behavior samples N, that is, the total number of interaction behaviors of all recorded teaching interaction behaviors and learning interaction behaviors. For each teacher, the present disclosure can also count the total number of teaching interaction behaviors Nt of the teacher, that is, the number of times T behavior appears. At the same time, the present disclosure can also traverse the character string composed of the learning interaction behavior S and the teaching interaction behavior T, such as TSSSTS...STS, to find out how many groups there are, and the specific number of groups is the total number of continuous behaviors in the present disclosure.

[0075] The present disclosure can determine the teacher behavior occupancy rate and the teacher-student behavior conversion rate based on the total number of teaching interaction behaviors, the total number of interactive behaviors, and the total number of continuous behaviors according to the following formula.

[0076]

[0077]

[0078] Among them, Rt is the teacher behavior occupancy rate, Ch is the teacher-student behavior conversion rate, Nt is the total number of teaching interaction behaviors, N is the total number of interactive behaviors, and g is the total number of continuous behaviors.

[0079] In an optional embodiment of the present disclosure, referring to Table 4 and Table 5, in the cases of Teacher A and Teacher B, the present disclosure records 20 interactive behavior samples, that is, the total number of interactive behaviors N is 20. The total number of teacher interactive behaviors T Nt of Teacher A is 13, while the total number of teacher interactive behaviors T Nt of Teacher B is 16. This indicates that Teacher B's lectures and guidance time in class are relatively long.

[0080] In order to calculate the teacher-student behavior conversion rate Ch, the present disclosure also needs to determine the total number of teacher-student continuous behaviors g. This is done by traversing a string consisting of S and T letters, and each continuous S or T sequence is regarded as an independent behavior group. In the case of Teacher A, the present disclosure identified 8 groups of continuous behaviors, that is, the total number of continuous behaviors g is 8; and in the case of Teacher B, 6 groups were identified, that is, the total number of continuous behaviors g is 6.

[0081] Based on the above data, the present disclosure can calculate the Rt and Ch values ​​of each teacher: Teacher B's Rt value is 0.8 ( ), indicating that teacher behavior accounts for a large proportion of the class time; the Ch value is 0.25 ( ), indicating that there are relatively few transitions between teacher and student behaviors, and this is a lecture-based classroom.

[0082] Teacher A's Rt value is 0.65 ( ), which is relatively low; while the Ch value is 0.35 ( ), which is relatively high, indicating that the transition between teacher and student behaviors in the classroom is more frequent, and it belongs to a mixed classroom.

[0083] Through the ST analysis method, the present invention can intuitively display the interaction between teachers and students in the classroom, and based on the two key indicators of teacher behavior occupancy rate and teacher-student behavior conversion rate, quantitatively evaluate the teacher's classroom performance, providing valuable reference for teachers to improve their teaching methods.

[0084] Step 303, based on the teacher behavior occupancy rate and the teacher-student behavior conversion rate, determine the class type corresponding to the teacher.

[0085] In some embodiments, the classroom types in the present disclosure may include practice type, lecture type, dialogue type and mixed type.

[0086] If the teacher behavior occupancy rate is less than or equal to the first preset threshold, the class type corresponding to the teacher is determined to be a practice type; if the teacher behavior occupancy rate is greater than or equal to the second preset threshold, the class type corresponding to the teacher is determined to be a lecture type; if the teacher-student behavior conversion rate is greater than or equal to the third preset threshold, the class type corresponding to the teacher is determined to be a dialogue type; if the teacher behavior occupancy rate is greater than the first preset threshold and less than the second preset threshold, and the teacher-student behavior conversion rate is less than the third preset threshold, the class type corresponding to the teacher is determined to be a mixed type.

[0087] The first preset threshold, the second preset threshold, and the third preset threshold can be set based on actual needs and are not limited in the embodiments of the present disclosure. The present disclosure can specifically take the first preset threshold as 0.3, the second preset threshold as 0.7, and the third preset threshold as 0.4 as an example, and determine the class type according to the determined teacher behavior occupancy rate Rt and the teacher-student behavior conversion rate Ch.

[0088] Step 304, based on the class type, determine the teacher's classroom interaction score under the classroom interaction indicator.

[0089] In some embodiments, the present disclosure may select a corresponding classroom interaction scoring algorithm according to a determined classroom type and determine a corresponding classroom interaction score.

[0090] If the class type is a practice type, the teacher's classroom interaction score can be determined based on the teacher's behavior occupancy rate and the first preset threshold; if the class type is a lecture type, the teacher's classroom interaction score can be determined based on the teacher's behavior occupancy rate and the second preset threshold; if the class type is a dialogue type or a mixed type, the frequency of occurrence and duration of effective interactive behavior sequences can be determined based on learning interactive behaviors and teaching interactive behaviors, and the interactive behavior effectiveness score can be determined based on the frequency of occurrence and duration, and the teacher's classroom interaction score can be determined based on the interactive behavior effectiveness score.

[0091] If the teacher's corresponding class type is practice type, the teacher's classroom interaction score under the classroom interaction indicator is determined according to the following formula 3: Formula 3 in, Score class interactions, For teacher behavior share, is a first preset threshold; If the teacher's corresponding class type is a lecture type, the teacher's classroom interaction score under the classroom interaction indicator is determined according to the following formula 4: Formula 4 in, Score class interactions, For teacher behavior share, is a second preset threshold; If the class type is mixed or dialogue type, in order to avoid misidentification, the teacher's classroom interaction score under the classroom interaction indicator is determined according to the following formula 5:

[0092] Formula 5 in, is the classroom interaction score, E is the interactive behavior effectiveness score, S is the effective interactive behavior sequence, is the preset weight corresponding to the effective interactive behavior sequence, It is a preset adjustment parameter used to balance the impact of frequency and duration on classroom interaction scores. is the frequency of occurrence of effective interactive behavior sequence S, is the duration of the effective interactive behavior sequence S, The total duration of the first class.

[0093] The total score of classroom interaction can be set to 5. The score range of hybrid and dialogic classrooms is limited to 3 to 5 points, and the interactive behavior effectiveness score E obtained in hybrid and dialogic classrooms needs to satisfy 0<E≤1.

[0094] In an optional embodiment of the present disclosure, the present disclosure may specifically utilize data mining technology to mine effective interactive behavior sequences based on learning interactive behaviors and teaching interactive behaviors, thereby obtaining the occurrence frequency and duration of the effective interactive behavior sequences.

[0095] Specifically, the present disclosure can select 50 course videos from the course videos in the region, and use data mining to obtain a set of interactive behavior sequences that can effectively achieve students' deep thinking through expert scoring and ranking, as shown in Table 6:

[0096] Table 6 In summary, the present disclosure accurately identifies and analyzes the learning interaction behaviors and teaching interaction behaviors in the classroom, calculates the teacher's behavior occupancy rate and the teacher-student behavior conversion rate, and further determines the teacher's classroom type. Finally, combined with the classroom type, a comprehensive and accurate classroom interaction score is given to the teacher under the classroom interaction index. The classroom interaction score not only reflects the teacher's classroom management ability, but also reflects the degree of student participation in classroom interaction, providing a strong reference for improving teaching quality and classroom interaction effects.

[0097] As a possible implementation, Figure 4 The flowchart of another teaching assistance method shown in the figure, based on the above embodiment, the classroom indicator includes a classroom fluency indicator, and the specific process of determining the score of each teacher in the first classroom under multiple classroom indicators based on the audio information and behavior information in the first classroom includes the following steps: Step 401: obtain an audio feature vector corresponding to the audio information, and perform mapping processing on the audio feature vector to obtain a disfluency label corresponding to the audio feature vector.

[0098] In some embodiments, since the present disclosure can call an audio pre-training model (i.e., a Wav2Vec model), the audio information is processed frame by frame. The process obtains the representation of the audio signal through the forward propagation of the audio pre-training model, and gradually divides it into frames to ensure that the speech features of each time segment are analyzed, thereby obtaining the audio feature vector corresponding to the audio information.

[0099] Specifically, the present invention can preprocess audio information to obtain audio information to be processed, and the preprocessing includes format conversion, resampling and frame processing; use an audio pre-training model to extract features of the audio information to be processed to obtain an audio feature vector; use a linear layer to map each frame of audio feature vector to a non-smooth label space to obtain a non-smooth label corresponding to each frame of audio feature vector.

[0100] The non-smoothness labels corresponding to the non-smoothness events in the present disclosure may be specifically shown in Table 7.

[0101]

[0102] Table 7 The specific process of preprocessing the audio information in the present disclosure includes: converting the original audio signal into a format suitable for model input. That is, resampling the audio data to resample the audio signal to the target sampling rate. ; Use the window function to frame the audio data, the window time is milliseconds, the number of samples per frame is , the number of overlapping samples is After that, the Wav2Vec model is used to extract frame-by-frame features. The forward propagation of the Wav2Vec model converts each frame of audio data into Mapped to a high-dimensional audio feature vector : ,in Represents the forward propagation function of the Wav2Vec model.

[0103] After the Wav2Vec model outputs, the audio feature vector Stored in a data structure. The feature vector of each frame is in the form of: , where T is the total number of frames.

[0104] Using a linear layer to map each frame of audio feature vector to the disfluency label space, obtaining the disfluency label corresponding to each frame of audio feature vector specifically includes: using multiple disfluency label types in the disfluency label control, mapping each frame of audio feature vector to the disfluency label space, and obtaining the mapped output vector; converting the output vector into a probability distribution of the disfluency label; for the probability distribution of each frame of disfluency label, selecting the disfluency label with the largest probability as the final disfluency label.

[0105] Specifically, the present disclosure can map the audio feature vectors obtained frame by frame to the disfluency label space, and use a linear layer to project the audio feature vector of each frame into a defined label space (such as word repetition, extension, etc.), thereby determining the disfluency label corresponding to each frame.

[0106] Among them, the present disclosure can set a non-smooth label space , respectively: Repeat for the word, To extend, For blocking, For incomplete sentences, Insert word In addition, add a clean speech tag: For clean speech, there are 6 labels in total.

[0107] Use a linear layer to transform each frame of audio feature vector Mapped to label space. The mathematical form of the linear layer is: , where W is the weight matrix, is the bias term, is the output vector.

[0108] Use the Softmax function to convert the output vector Convert to probability distribution of disfluent labels: ,in Represents a given audio feature vector When not smooth label probability.

[0109] For each frame, choose the most likely label according to the probability distribution:

[0110] in, Yes Frame Predicted disfluency labels.

[0111] The present disclosure can also adapt the linear layer to a specific downstream task by fine-tuning the linear layer. During the fine-tuning, the performance of the model in the recognition of non-smooth events is gradually optimized, and hyperparameters such as the learning rate and batch size are adjusted to prevent overfitting, and the best model is selected based on the performance on the validation set.

[0112] It is understandable that, since the present disclosure needs to utilize an audio pre-training model to process audio information, the present disclosure needs to perform pre-training processing on the audio pre-training model.

[0113] Among them, the audio pre-training model disclosed in the present invention can use the self-supervised learning framework Wav2Vec model to extract the structural features of speech and improve the generalization ability of the model.

[0114] After that, the present disclosure can construct a pre-trained annotated audio data set to determine the disfluency label corresponding to the audio feature vector of the audio information using the pre-trained annotated audio data set. Specifically, accurate text transcription can be generated for each audio clip through manual annotation, including event types (such as word repetitions, pauses, etc.) and timestamps (start and end time), and disfluency events in the audio can be analyzed word by word, and different types of events can be annotated with predefined labels.

[0115] The labeling process needs to ensure accurate alignment between phonetic characters (the labeled data set will be used for subsequent model fine-tuning). In particular, accent and noise data need special processing to ensure that the model can handle complex audio environments.

[0116] For example: "We can / rwe can / rgo first / pgo first / reat / p, and then / b".

[0117] For example: "I think this poem expresses the poet's / b er / b er, lofty patriotic ideas."

[0118] In addition, before the present disclosure uses the audio pre-training model to process the audio information, the present disclosure also needs to pre-process the collected audio data, that is, apply the noise removal algorithm to denoise. At the same time, data enhancement technology (pitch and speed adjustment) is used to expand the diversity of the data set, and then Kaldi is used to segment the speech, generate the transcription of the speech segment and the audio alignment result, and improve the data quality and accuracy.

[0119] Step 402: for each jerky event corresponding to the jerky label of each frame, determine the occurrence frequency of each jerky event.

[0120] In some embodiments, after obtaining the disfluency label corresponding to each frame of audio feature vector, the present disclosure can count the disfluency events in the audio information, including the frequency and duration of each disfluency, and construct a time distribution graph of classroom disfluency events.

[0121] Specifically, in order to improve the efficiency and accuracy of statistics, the present disclosure can divide the audio information into multiple windows according to a preset step size, the window size is B frame, the sliding step size is S frame, and the audio information is divided into multiple small segments by sliding the window. For each frame f in the sliding window, the occurrence frequency of all non-smooth events is counted ,in represents the disfluency event corresponding to the disfluency label k.

[0122] For frame f, it belongs to the non-smooth label Case: Define an indicator function , when the jth frame belongs to the unsmooth label hour, =1 if the value is set to 0, otherwise it is 0.

[0123] For the i-th sliding window , determine whether there are all non-smooth events: >0 Let M be the total number of sliding windows, and the formula for summing the number of windows with unsmooth labels in all windows is:

[0124] Based on the frame-by-frame statistical results, a time distribution diagram of classroom disfluency events is generated. Figure 5 As shown, the present disclosure provides an event distribution diagram of a specific non-fluent event. Figure 5 The time distribution of each disfluency event in the class and its frequency and duration are shown. Figure 5 , there are T audio window frames, corresponding to K disfluent labels, and we can get a matrix ,in It means that the kth type of non-smooth event occurred in the tth frame.

[0125] Step 403: Based on the occurrence frequency of each disfluency event, determine the negative impact score of each disfluency event on class fluency.

[0126] In some embodiments, the overall fluency of the class is calculated based on the importance of each disfluency event (for example, word repetition and speech speed change may have different weights), combined with the occurrence frequency and duration of the disfluency events determined above. The final fluency score can be generated by weighted average to reflect the fluency of the class performance.

[0127] Each disfluency event has a predefined weight .

[0128] is the weight set of non-smooth events, and .

[0129] Weighted calculation of the impact of each disfluency event: Each disfluency label , combined with its frequency of occurrence , is the total number of windows of audio information, and its negative impact on class fluency is calculated.

[0130] Specifically, the negative impact score of each disfluency event on class fluency can be determined based on the frequency of occurrence of each disfluency event according to the following formula 6: Formula 6 in, is the negative impact score of the disfluency event Ck on the fluency of the class, Ck is the disfluency event corresponding to the disfluency label k, is the preset weight corresponding to the non-smooth label k, It is a preset adjustment parameter used to adjust the negative impact of the frequency of occurrence of non-smooth events. is the occurrence frequency of the non-smooth event Ck, is the total number of windows corresponding to the audio information.

[0131] Step 404: Determine the teacher's class fluency score under the class fluency index based on the negative impact score.

[0132] In some embodiments, the teacher's class fluency score under the class fluency index may be determined based on the negative impact score according to the following formula 7: Formula 7 in, Score the fluency of the class, T is the total score of the preset threshold, Score negative impact.

[0133] Among them, when the duration of the non-fluidity event is longer, the negative impact score will be larger and the class fluency score will be lower. The closer the class fluency score is to the preset threshold total score, the smoother the class is; conversely, the closer the class fluency score is to 0, the less smooth the class is.

[0134] In summary, this disclosure can accurately identify and mark disfluency events by deeply analyzing the feature vectors in the audio information, and then count the frequency of occurrence of each disfluency event. Based on the frequency of occurrence, the negative impact of each disfluency event on classroom fluency is further evaluated, and finally a comprehensive score is given to the teacher on the key indicator of classroom fluency, which not only reflects the fluency of the teacher in the teaching process, but also provides a valuable reference for improving teaching quality and student learning experience.

[0135] As a possible implementation, Figure 6 The flowchart of another teaching assistance method shown in the figure, based on the above embodiment, determines whether a teacher is a teacher to be assisted according to the teacher's scores under multiple classroom indicators, including the following steps: Step 501, statistics the teacher's scores under different classroom indicators to obtain the teacher's comprehensive score.

[0136] In some embodiments, after obtaining the teacher's scores under different classroom indicators, that is, after obtaining the teacher's classroom content quality score, classroom interaction score and classroom fluency score, the present disclosure can add the classroom content quality score, classroom interaction score and classroom fluency score to obtain the teacher's comprehensive score.

[0137] Among them, the present disclosure can directly add the class content quality score, the class interaction score and the class fluency score to obtain a comprehensive score, as shown in the following formula:

[0138] in, Rate the quality of the course content, Score class interactions, Score class fluency.

[0139] The present disclosure can also set different scoring weights for different classroom indicators, and add the classroom content quality score, classroom interaction score, and classroom fluency score based on the scoring weights to obtain a comprehensive score, as shown in the following formula:

[0140] in, Rate the quality of the course content, The scoring weight corresponding to the quality of the classroom content, Score class interactions, The scoring weight corresponding to the classroom interaction score, Score class fluency, The scoring weight corresponding to the class fluency score.

[0141] It can be understood that the scoring weights corresponding to different classroom indicators in the present disclosure can be set according to actual conditions and are not limited in the embodiments of the present disclosure.

[0142] Step 502: If the comprehensive score and the score results corresponding to each classroom indicator under different classroom indicators do not meet the expected conditions, the teacher is determined to be a teacher to be assisted. The expected conditions are that the comprehensive score is greater than or equal to the comprehensive threshold, and the score results corresponding to each classroom indicator under different classroom indicators are greater than or equal to the score threshold corresponding to each classroom indicator.

[0143] In some embodiments, the present disclosure defines the judgment rules for the object to be assisted (teachers to be assisted) and the final criteria for not being assisted based on the predetermined expected conditions. This is intended to help the teaching team quantify the rules for forming assistance behaviors, and through the dynamic optimization of the assistance scheme proposed in the present disclosure, it is ultimately ensured that the assisted object can complete the assistance indicators and form a closed loop of assistance actions.

[0144] Here, the multiple classroom indicators disclosed in the present invention are classroom content quality score, classroom interaction score and classroom fluency score, and the continuous score range of each classroom indicator is 0-5.

[0145] The desired conditions of the present disclosure are as follows: When the teacher's comprehensive score meets the comprehensive threshold and the score results corresponding to each classroom indicator under different classroom indicators are greater than or equal to the score threshold corresponding to each classroom indicator, it is considered that the teacher does not need assistance and meets the expected conditions; otherwise, it is considered that the current teacher needs assistance, that is, the current teacher is a teacher to be assisted.

[0146] In summary, the present invention uses pre-defined assistance conditions combined with the teacher's scores under various classroom indicators to accurately determine the teachers who need assistance, so as to achieve precise assistance for teachers.

[0147] As a possible implementation, Figure 7 The flowchart of another teaching assistance method shown in the figure, based on the above embodiment, if the teacher is a teacher to be assisted, the specific process of determining the recommended famous teacher corresponding to the teacher to be assisted from the famous teacher database based on the scores of the teacher to be assisted under multiple classroom indicators includes the following steps: Step 601, determine the classroom support indicators and support intensity corresponding to the teacher to be supported.

[0148] In some embodiments, the present disclosure may determine the classroom assistance indicators and assistance intensity of the teacher to be assisted based on the scores of the teacher to be assisted under various classroom indicators.

[0149] When the comprehensive score of the teacher to be assisted meets the comprehensive threshold, but the score of some classroom indicators is lower than the corresponding score threshold, it is considered that the teacher has a large distribution deviation and needs to be assisted in the weak dimension. The classroom indicators with scores lower than the score threshold are selected as the assisted classroom indicators. When the comprehensive score of the teacher to be assisted does not meet the comprehensive threshold, but the score of some classroom indicators is greater than the corresponding score threshold, it is considered that the teacher has a strong deviation between dimensions and needs to be assisted in a focused manner on the weak dimensions. Classroom indicators other than those with scores greater than the corresponding score threshold are selected as assisted classroom indicators. When the comprehensive score of the teacher to be assisted does not meet the comprehensive threshold and the score under each classroom indicator is less than the corresponding score threshold, it is considered that the teacher needs assistance under each classroom indicator, and all classroom indicators are used as assistance classroom indicators.

[0150] At the same time, after determining the teachers to be assisted, the present disclosure can mark the intensity of assistance for the teachers to be assisted according to the difference in the scores of the teachers' various classroom indicators and the number of assisted classroom indicators, and classify the intensity of assistance for the teachers to be assisted. The intensity of assistance classification includes three categories: Strong assistance, Medium assistance, and Normal assistance. This classification is used to map the overall distribution of the learning curve of the group of teachers to be assisted, and assist in the subsequent dynamic adjustment of the learning curve adjustment amplitude factor of the current teachers to be assisted. , and find famous teachers with more appropriate intensity to support them.

[0151] To this end, this disclosure uses the cumulative difference method to reflect the difference between the current teachers to be assisted and the dimension threshold, and to show the real difference between the current teachers to be assisted and the compliance standard. At the same time, since the cumulative difference value of the teachers to be assisted follows a normal distribution, the current assistance intensity label range definition is measured according to the group difference distribution.

[0152] The present disclosure can determine the cumulative difference corresponding to each classroom indicator of the teacher to be assisted through the following formula:

[0153] in, is the score difference for each classroom indicator i, For each class indicator, The scoring threshold corresponding to each classroom indicator, The cumulative difference corresponding to each classroom indicator.

[0154] The cumulative difference corresponding to each classroom indicator of the teacher to be assisted The smaller it is, the weaker the support the current teachers in need of support are; conversely, the stronger the support the current teachers in need of support need.

[0155] like Figure 8 The cumulative difference distribution diagram of the teachers to be assisted is shown in FIG. The cumulative difference distribution of the teachers to be assisted can be expressed as follows:

[0156] , represents the overall cumulative difference mean , represents the overall standard deviation definition (Strong support) Range: SumD<

[0157] definition (Medium assistance) Scope: ≤SumD≤

[0158] definition (Weak support) Range: SumD> .

[0159] Step 602: Using the support intensity, support classroom indicators and a preset scoring system, determine the support expectations of each famous teacher in the famous teacher database.

[0160] In some embodiments, the present disclosure can use the ELO scoring system (i.e., a preset scoring system) to complete the calculation of the expected support score of the famous teacher for the teacher under the support classroom indicator. The higher the score of the expected support score, the higher the support matching degree of the famous teacher for the supported teacher.

[0161] Specifically, the present disclosure can determine the expected score of assistance given to teachers by each famous teacher in the famous teacher database according to the following formula 8, using the assistance intensity, the assistance classroom index, and the preset scoring system: Formula 8 in, is the score of famous teacher M under the supporting classroom indicator i, is the score of teacher N under the support classroom indicator i; is the score of the famous teacher M's expectation of helping teachers, k is the preset factor, is the support weight of supporting classroom indicator i.

[0162] Step 603: Based on the expected support score and the preset effective range of the support learning curve, the famous teachers in the famous teacher database are matched and filtered to obtain a plurality of pre-recommended famous teachers.

[0163] In some embodiments, the Elo rating system can help match the current teacher to be assisted with a famous teacher who is suitable for the difficulty level. By combining the famous teacher's assistance with the assistance intensity classification of the teacher to be assisted, the effective range of the assistance learning curve (i.e., the effective range of the famous teacher's assistance learning curve) is set according to the intensity classification of the teacher to be assisted. ), the effective interval of the preset assistance learning curve is dynamically adjusted based on the assistance effect of famous teachers and the overall distribution changes of the classification group (dynamic adjustment parameters: ). Through interval setting, the disclosure can prevent overfitting of similar famous teachers and ensure that the final famous teacher pre-recommendation has a certain learning curve.

[0164] Combining the ELO scoring system and famous teachers to help the effective range of the learning curve , the present disclosure can realize personalized recommendation of famous teachers for supporting teachers through the above formula 8. Support expectation score The higher the value, the higher the current famous teacher's expectations for the support of the supporting teachers, and the more obvious the support effect; conversely, the lower the current famous teacher's expectations for the support of the supporting teachers, the more ordinary the support effect.

[0165] According to the support expectation scores of each famous teacher for the support teachers under the support classroom indicator Effective range of learning curve , filter the famous teachers by matching degree and get the pre-recommended famous teachers.

[0166] in, .

[0167] The effective interval of the preset assistance learning curve for the teachers to be assisted is defined as ; .

[0168] The larger it is, the more difficult it is for teachers to improve; conversely, the smaller it is, the easier it is for teachers to improve.

[0169] Initial iteration, this paper defines the effective interval of the assistance learning curve : The support intensity of the teachers to be supported is classified as follows: :definition , it is believed that the current assistance learning curve is more suitable for the current objects to be assisted in a smoother range; The support intensity of the teachers to be supported is classified as follows: :definition ,Thinking that the current assistance learning curve is more suitable for the current teachers to be assisted with a certain degree of improvement; The support intensity of the teachers to be supported is classified as follows: :definition ,Thinking that the current assistance learning curve is more suitable for the current teachers to be assisted under a stronger degree of improvement; In subsequent iterations, this disclosure can be used to improve the actual learning level of the teachers to be assisted under the current learning curve. And the learning improvement degree of the support intensity classification of the teacher to be supported is dynamically adjusted, for example Fig. 9 Schematic diagram of the changes in the assistance learning curve shown.

[0170] The indicators of the supported classes that need support must be met That is, the learning curve of the famous teacher in the support dimension must meet the specified preset support learning curve effective range Inside.

[0171] And other non-support classroom indicators need to meet , which means that the score of the famous teacher under the non-support classroom indicator must exceed the score threshold .

[0172] This publication is based on the scores of each famous teacher's expectations for the support given to teachers. Effective range of learning curve The famous teachers who meet the personalized assistance recommendation will be screened and their expectations for assistance will be scored. Sorting. The degree of assistance provided by famous teachers will be ranked according to the expected assistance score, and a preset number of famous teachers will be selected as pre-recommended famous teachers. For example, the top 5 famous teachers will be selected as pre-recommended famous teachers to make targeted recommendations to the teachers to be assisted. .

[0173] Step 604: Determine the degree of compatibility between the teacher to be assisted and a plurality of pre-recommended famous teachers, and based on the degree of compatibility, select a recommended famous teacher corresponding to the teacher from the plurality of pre-recommended famous teachers.

[0174] In some embodiments, by obtaining pre-recommended famous teachers, the present disclosure can calculate the scores of the teachers to be assisted under various classroom indicators after the assistance. Using the fitness evaluation, the real effect of the assistance of the current pre-recommended famous teachers on the improvement of the assisted teachers can be truly reflected. The higher the fitness score, the more effective the actual assistance of the pre-recommended famous teacher to the assisted teachers is, and the support of the pre-recommended famous teacher to the assisted teachers will be emphasized. Through the measurement of fitness, the present disclosure can explore the potential demands of the teachers to be assisted for the assisting famous teachers, and under the dual recommendation of the explicit dimension and the latent dimension, more suitable assistance from famous teachers can be inferred.

[0175] Among them, the present disclosure defines that the ratio of the improvement scores of the current teachers to be assisted in various classroom indicators after assistance and the expected improvement scores of the pre-recommended famous teachers for the assisted teachers can reflect the improvement degree of the current teachers to be assisted. , as shown in the following formula:

[0176] in, is the score of the teacher to be assisted under the i-th classroom indicator, is the score of the teacher to be assisted under the i-th classroom indicator after assistance, is the score of the pre-recommended master teacher under the i-th classroom indicator.

[0177] This public definition of the suitability of pre-recommended famous teachers for supporting teachers The difference between the actual improvement and the expected support score for each classroom indicator is summed to reflect the actual support results of the current pre-recommended famous teachers in treating the support teachers under the support classroom indicators. The adaptability is calculated by the following formula:

[0178] in, Fit: measure the actual assistance results of famous teacher M to teacher N; is the support weight of the i-th classroom indicator, is the expected support score for the pre-recommended famous teacher M under the i-th classroom indicator, The degree of improvement of teacher N after assistance.

[0179] According to the fitness degree corresponding to each pre-recommended master teacher, the master teacher who best matches the current teacher to be assisted can be determined, and the pre-recommended master teacher can be used as the recommended master teacher, aiming to optimize the fitness degree of the master teacher database and prune it.

[0180] At the same time, after obtaining the improvement degree of the current teacher to be assisted under the assistance of the famous teacher, the present disclosure can also calculate the effective interval of the assistance learning curve of the teacher to be assisted. According to the current support intensity of the teacher to be supported, the corresponding improvement distribution is obtained:

[0181]

[0182] According to the degree of improvement of recommended famous teachers towards supporting teachers Distribution interval, to help update and optimize the effective interval of the learning curve:

[0183]

[0184] Adjust parameters for the valid period of the learning curve.

[0185] This disclosure can reflect the actual learning situation of the current teachers to be assisted in the subsequent personalized matching adjustment of famous teachers through personalized dynamic adjustment of the effective interval of the assistance learning curve of the teachers to be assisted. It ensures that the teachers to be assisted with potential learning can be further improved in the subsequent assistance plan; it ensures that the teachers who need more smooth assistance can get a more comfortable improvement in the subsequent assistance, and truly realizes that personalized assistance follows the actual growth characteristics of the assisted individuals.

[0186] After determining the recommended famous teacher corresponding to the current teacher to be assisted, the disclosure can repeat the above content for other teachers in the first classroom, calculate the scores of each teacher under multiple classroom indicators, and provide famous teacher assistance to the teacher to be assisted according to the scores. If all the teachers to be assisted in the current first classroom meet the expected conditions, the assistance will be stopped; otherwise, the recommendation optimization will be performed based on the scores of the current teachers to be assisted to ensure that all teachers to be assisted can complete the closed loop from needing assistance to meeting the goals.

[0187] In summary, the present disclosure calculates the expected score of teachers' assistance for each famous teacher by accurately measuring the assistance needs and assistance intensity of the teachers to be assisted, combining the assistance classroom indicators and scoring system. Subsequently, according to these expected scores of assistance and the effective interval of the preset assistance learning curve, multiple pre-recommended famous teachers matching the teachers to be assisted are screened out. Finally, by further evaluating the compatibility between the teachers to be assisted and these pre-recommended famous teachers, the most suitable recommended famous teachers are selected to ensure the maximization of the assistance effect.

[0188] Corresponding to the above teaching assistance method, the present invention also provides a teaching assistance device. Since the device embodiment of the present invention corresponds to the above method embodiment, details not disclosed in the device embodiment can be referred to the above method embodiment, and will not be repeated in the present invention.

[0189] Fig.10 A structural diagram of a teaching assistance device provided by an embodiment of the present disclosure is shown in FIG. Fig.10 As shown, the device comprises: The scoring unit 1010 is used to determine the scores of the teacher in the first class under multiple classroom indicators based on the audio information and behavior information in the first class; A screening unit 1020 is used to determine whether a teacher is a teacher to be assisted based on the teacher's scores under multiple classroom indicators; The famous teacher recommendation unit 1030 is used to determine the recommended famous teacher corresponding to the teacher to be assisted from the famous teacher database based on the scores of the teacher to be assisted under multiple classroom indicators if the teacher is the teacher to be assisted, and recommend the recommended famous teacher to the teacher to be assisted, so as to use the recommended famous teacher to provide teaching assistance to the teacher to be assisted.

[0190] In some embodiments of the present disclosure, the plurality of classroom indicators include a classroom content quality indicator, and the scoring unit 1010 is used to: Determine the scores of the students' in-class tests corresponding to the first class; Obtain classroom performance scores corresponding to behavioral information; Based on students' in-class test scores and classroom performance scores, determine the teacher's classroom content quality score under the classroom content quality index.

[0191] In some embodiments of the present disclosure, the scoring unit 1010 is used to: According to the following formula 1, the scores of the students' in-class test corresponding to the first class are determined:

[0192]

[0193] Formula 1 in, is the relative score of the class test of the i-th student, is the absolute score of the class test of the i-th student, is the mean of the absolute scores of the students’ in-class tests, is the standard deviation of the students’ absolute scores in the in-class tests, N is the number of students, Q is the mean of the students’ relative scores in the in-class tests, Scoring students' in-class tests, is the preset weight, The default basic score is: According to the following formula 2, based on the students' in-class test scores and classroom performance scores, the teacher's classroom content quality score under the classroom content quality indicator is determined: Formula 2 in, Rate the quality of the course content, Scoring students' in-class tests, The preset weights corresponding to the students' in-class test scores, Grade class performance. Preset weights for grading class performance.

[0194] In some embodiments of the present disclosure, the plurality of classroom indicators include a classroom interaction indicator, and the scoring unit 1010 is used to: Determine the learning interaction behavior and teaching interaction behavior corresponding to the behavior information; Based on learning interaction behaviors and teaching interaction behaviors, determine the teacher behavior occupancy rate and teacher-student behavior conversion rate; Based on the teacher behavior occupancy rate and the teacher-student behavior conversion rate, determine the class type corresponding to the teacher; Based on the class type, determine the teacher's classroom interaction score under the classroom interaction indicator.

[0195] In some embodiments of the present disclosure, the scoring unit 1010 is used to: If the teacher behavior occupancy rate is less than or equal to the first preset threshold, the class type corresponding to the teacher is determined to be a practice type, and the teacher's classroom interaction score under the classroom interaction index is determined according to the following formula 3: Formula 3 in, Score class interactions, For teacher behavior share, is a first preset threshold; If the teacher behavior occupancy rate is greater than or equal to the second preset threshold, the class type corresponding to the teacher is determined to be a lecture type, and the classroom interaction score of the teacher under the classroom interaction index is determined according to the following formula 4: Formula 4 in, Score class interactions, For teacher behavior share, is a second preset threshold; If the teacher-student behavior conversion rate is greater than or equal to the third preset threshold, the class type corresponding to the teacher is determined to be a dialogue type. If the teacher behavior occupancy rate is greater than the first preset threshold and less than the second preset threshold, and the teacher-student behavior conversion rate is less than the third preset threshold, the class type corresponding to the teacher is determined to be a mixed type. According to the following formula 5, the classroom interaction score of the teacher under the classroom interaction index is determined:

[0196] Formula 5 in, is the classroom interaction score, E is the interactive behavior effectiveness score, S is the effective interactive behavior sequence, is the preset weight corresponding to the effective interactive behavior sequence, To preset adjustment parameters, is the frequency of occurrence of effective interactive behavior sequence S, is the duration of the effective interactive behavior sequence S, The total duration of the first class.

[0197] In some embodiments of the present disclosure, the classroom index includes a classroom fluency index, and the scoring unit 1010 is used to: Obtaining an audio feature vector corresponding to the audio information, and performing mapping processing on the audio feature vector to obtain a disfluency label corresponding to the audio feature vector; For the non-smooth events corresponding to the non-smooth labels of each frame, determine the occurrence frequency of each non-smooth event; Based on the frequency of each disfluency event, determine the negative impact score of each disfluency event on class fluency; Based on the negative impact score, determine the teacher's classroom fluency score under the classroom fluency index.

[0198] In some embodiments of the present disclosure, the scoring unit 1010 is used to: According to the following formula 6, the negative impact score of each disfluency event on class fluency is determined based on the frequency of occurrence of each disfluency event: Formula 6 in, is the negative impact score of the disfluency event Ck on the fluency of the class, Ck is the disfluency event corresponding to the disfluency label k, is the preset weight corresponding to the non-smooth label k, To preset adjustment parameters, is the occurrence frequency of the non-smooth event Ck, is the total number of windows corresponding to the audio information; Based on the negative impact score, the teacher's class fluency score under the class fluency index is determined according to the following formula 7: Formula 7 in, Score the fluency of the class, T is the total score of the preset threshold, Score negative impact.

[0199] In some embodiments of the present disclosure, the screening unit 1020 is used to: The teacher's scores under different classroom indicators are counted to obtain the teacher's comprehensive score; If the comprehensive score and the scoring results corresponding to each classroom indicator under different classroom indicators do not meet the expected conditions, the teacher will be determined as a teacher to be assisted. The expected conditions are that the comprehensive score is greater than or equal to the comprehensive threshold, and the scoring results corresponding to each classroom indicator under different classroom indicators are greater than or equal to the scoring threshold corresponding to each classroom indicator.

[0200] In some embodiments of the present disclosure, the famous teacher recommendation unit 1030 is used to: Determine the classroom support indicators and support intensity corresponding to the teachers to be supported; Using the support intensity, support classroom indicators and the preset scoring system, determine the support expectations of each famous teacher in the famous teacher database. Based on the expected support score and the effective range of the preset support learning curve, the famous teachers in the famous teacher database are matched and filtered to obtain multiple pre-recommended famous teachers; Determine the compatibility between the teacher to be assisted and multiple pre-recommended famous teachers, and based on the compatibility, select the recommended famous teacher corresponding to the teacher from multiple pre-recommended famous teachers.

[0201] In some embodiments of the present disclosure, the famous teacher recommendation unit 1030 is used to: According to the following formula 8, using the support intensity, support classroom indicators and the preset scoring system, the expected support score of each famous teacher in the famous teacher database for teachers is determined: Formula 8 in, is the score of famous teacher M under the supporting classroom indicator i, is the score of teacher N under the support classroom indicator i; is the score of the famous teacher M's expectation of helping teachers, k is the preset factor, is the support weight of supporting classroom indicator i.

[0202] It should be noted that the above explanation of the method embodiment is also applicable to the device of this embodiment, and the principle is the same, which is not limited in this embodiment.

[0203] Based on the above Figures 1 to 9 The method shown in the figure, accordingly, this embodiment also provides a computer program product, including a computer program, which implements the above-mentioned Figures 1 to 9 The method shown.

[0204] Based on the above Figures 1 to 9 The method shown in the embodiment also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the above-mentioned Figures 1 to 9 The method shown.

[0205] Based on this understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, USB flash drive, mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods of various implementation scenarios of the present application.

[0206] like Fig.11 The figure shows a hardware structure diagram of an electronic device of the present invention, including: at least one processor 1101; and, A memory 1102 is communicatively connected to at least one of the processors 1101; wherein: The memory 1102 stores instructions that can be executed by at least one of the processors. The instructions are executed by at least one of the processors so that the at least one of the processors can execute the teaching assistance method as described above.

[0207] Fig.11 A processor 1101 is taken as an example.

[0208] The electronic device may further include: an input device 1103 and a display device 1104 .

[0209] The processor 1101, the memory 1102, the input device 1103 and the display device 1104 may be connected via a bus or other means, and the figure takes the connection via a bus as an example.

[0210] The memory 1102 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as program instructions / modules corresponding to the review content generation method in the embodiment of the present application, for example, Figures 1 to 9 The processor 1101 executes various functional applications and data processing by running the non-volatile software programs, instructions and modules stored in the memory 1102, that is, the teaching assistance method in the above embodiment is implemented.

[0211] The memory 1102 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the review content generation method, etc. In addition, the memory 1102 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage devices. In some embodiments, the memory 1102 may optionally include a memory remotely arranged relative to the processor 1101, and these remote memories may be connected to the device for executing the review content generation method via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0212] The input device 1103 may receive user clicks and generate signal inputs related to user settings and function controls of the review content generation method. The display device 1104 may include display devices such as a display screen.

[0213] The one or more modules are stored in the memory 1102 and, when executed by the one or more processors 1101, execute the teaching assistance method in any of the above method embodiments.

[0214] Optionally, the above-mentioned physical device may also include a user interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a WI-FI module, etc. The user interface may include a display, an input unit such as a keyboard, etc., and the optional user interface may also include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a WI-FI interface), etc.

[0215] Those skilled in the art will appreciate that the above-mentioned physical device structure provided in this embodiment does not constitute a limitation on the physical device, and may include more or fewer components, or a combination of certain components, or different arrangements of components.

[0216] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the hardware and software resources of the above-mentioned physical device, and supports the operation of the information processing program and other software and / or programs. The network communication module is used to realize the communication between the components inside the storage medium, and the communication with other hardware and software in the information processing physical device.

[0217] Through the description of the above implementation methods, the technicians in this field can clearly understand that the present application can be implemented by means of software plus the necessary general hardware platform, or by hardware. By applying the scheme of this embodiment, compared with the current prior art, this embodiment determines the score of the teacher in the first classroom under multiple classroom indicators based on the audio information and behavior information in the first classroom; determines whether the teacher is a teacher to be assisted based on the teacher's score under multiple classroom indicators; if the teacher is a teacher to be assisted, then based on the score of the teacher to be assisted under multiple classroom indicators, the recommended famous teacher corresponding to the teacher to be assisted is determined from the famous teacher database, and the recommended famous teacher is recommended to the teacher to be assisted, so as to use the recommended famous teacher to provide teaching assistance to the teacher to be assisted, realize the objective and comprehensive evaluation of the teacher's teaching level, and provide accurate famous teacher recommendations for teachers who need assistance, which is helpful to improve the quality and teaching level of the overall teaching team, and provide strong support for improving teaching quality and promoting teaching innovation.

[0218] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0219] The above is only a specific implementation of the present application, so that those skilled in the art can understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and novel features applied for herein.

Claims

1. A teaching assistance method, characterized in that: The method comprises: Based on the audio information and behavior information in the first classroom, determining the scores of the teacher in the first classroom under multiple classroom indicators; Determine whether the teacher is a teacher to be assisted based on the teacher's scores under multiple classroom indicators; If the teacher is a teacher to be assisted, then based on the scores of the teacher to be assisted under the multiple classroom indicators, the recommended famous teacher corresponding to the teacher to be assisted is determined from the famous teacher database, and the recommended famous teacher is recommended to the teacher to be assisted, so as to use the recommended famous teacher to provide teaching assistance to the teacher to be assisted.

2. The method according to claim 1, characterized in that The multiple classroom indicators include a classroom content quality indicator, and the step of determining the score of each teacher in the first classroom under the multiple classroom indicators based on the audio information and behavior information in the first classroom includes: Determine the scores of the students' in-class test corresponding to the first class; Obtaining a classroom performance score corresponding to the behavior information; Based on the student's in-class test score and the classroom performance score, determine the teacher's classroom content quality score under the classroom content quality index.

3. The method according to claim 2, characterized in that Determining the scores of the student in-class test corresponding to the first class includes: The student in-class test score corresponding to the first class is determined according to the following formula 1: Formula 1 in, is the relative score of the class test of the i-th student, is the absolute score of the class test of the i-th student, is the mean of the absolute scores of the students’ in-class tests, is the standard deviation of the students’ absolute scores in the in-class tests, N is the number of students, Q is the mean of the students’ relative scores in the in-class tests, Scoring students' in-class tests, is the preset weight, The default basic score is: According to the following formula 2, based on the student's in-class test score and the classroom performance score, the teacher's classroom content quality score under the classroom content quality indicator is determined: Formula 2 in, Rate the quality of the course content, Scoring students' in-class tests, is the preset weight corresponding to the student's in-class test score, Grade class performance. The preset weight corresponding to the classroom performance score.

4. The method according to claim 1, characterized in that: The multiple classroom indicators include a classroom interaction indicator, and determining the score of each teacher in the first classroom under the multiple classroom indicators based on the audio information and behavior information in the first classroom includes: Determine the learning interaction behavior and the teaching interaction behavior corresponding to the behavior information; Based on the learning interaction behavior and the teaching interaction behavior, determine the teacher behavior occupancy rate and the teacher-student behavior conversion rate; Determine the class type corresponding to the teacher based on the teacher behavior occupancy rate and the teacher-student behavior conversion rate; Based on the class type, determine the teacher's classroom interaction score under the classroom interaction indicator.

5. The method according to claim 4, characterized in that Determining the classroom interaction score of the teacher under the classroom interaction indicator based on the teacher behavior occupancy rate and the teacher-student behavior conversion rate includes: If the teacher behavior occupancy rate is less than or equal to the first preset threshold, the class type corresponding to the teacher is determined to be a practice type, and the class interaction score of the teacher under the class interaction index is determined according to the following formula 3: Formula 3 in, Score class interactions, For teacher behavior share, is a first preset threshold; If the teacher behavior occupancy rate is greater than or equal to the second preset threshold, the class type corresponding to the teacher is determined to be a lecture type, and the classroom interaction score of the teacher under the classroom interaction index is determined according to the following formula 4: Equation 4 in, Score class interactions, For teacher behavior share, is a second preset threshold; If the teacher-student behavior conversion rate is greater than or equal to the third preset threshold, the class type corresponding to the teacher is determined to be a conversational type. If the teacher behavior occupancy rate is greater than the first preset threshold and less than the second preset threshold, and the teacher-student behavior conversion rate is less than the third preset threshold, the class type corresponding to the teacher is determined to be a mixed type. The classroom interaction score of the teacher under the classroom interaction index is determined according to the following formula 5: Formula 5 in, is the classroom interaction score, E is the effectiveness score of the interaction behavior, S is the effective interaction behavior sequence, is the preset weight corresponding to the effective interactive behavior sequence, To preset adjustment parameters, is the frequency of occurrence of effective interactive behavior sequence S, is the duration of the effective interactive behavior sequence S, The total duration of the first class.

6. The method according to claim 1, characterized in that The classroom indicator includes a classroom fluency indicator, and determining the score of each teacher in the first classroom under multiple classroom indicators based on the audio information and behavior information in the first classroom includes: Acquire an audio feature vector corresponding to the audio information, and perform mapping processing on the audio feature vector to obtain a disfluency label corresponding to the audio feature vector; For the non-smooth events corresponding to the non-smooth labels of each frame, determine the occurrence frequency of each non-smooth event; Based on the occurrence frequency of each disfluency event, determining a negative impact score of each disfluency event on class fluency; Based on the negative impact score, a classroom fluency score of the teacher under a classroom fluency index is determined.

7. The method according to claim 6, characterized in that Determining the negative impact score of each non-fluency event on class fluency based on the occurrence frequency of each non-fluency event includes: According to the following formula 6, based on the occurrence frequency of each non-fluency event, the negative impact score of each non-fluency event on class fluency is determined: Formula 6 in, is the negative impact score of the disfluency event Ck on the fluency of the class, Ck is the disfluency event corresponding to the disfluency label k, is the preset weight corresponding to the non-smooth label k, To preset adjustment parameters, is the occurrence frequency of the non-smooth event Ck, is the total number of windows corresponding to the audio information; According to the following formula 7, based on the negative impact score, the classroom fluency score of the teacher under the classroom fluency index is determined: Formula 7 in, is the fluency score of the class, T is the total score of the preset threshold, Score the negative impact.

8. The method according to claim 1, characterized in that Determining whether the teacher is a teacher to be assisted based on the teacher's scores under multiple classroom indicators includes: Counting the scores of the teacher under different classroom indicators to obtain a comprehensive score of the teacher; If the comprehensive score and the scoring results corresponding to each classroom indicator under the different classroom indicators do not meet the expected conditions, the teacher is determined to be a teacher to be assisted. The expected conditions are that the comprehensive score is greater than or equal to the comprehensive threshold, and the scoring results corresponding to each classroom indicator under the different classroom indicators are greater than or equal to the scoring threshold corresponding to each classroom indicator.

9. The method according to claim 1, characterized in that: If the teacher is a teacher to be assisted, then based on the scores of the teacher to be assisted under the multiple classroom indicators, a recommended famous teacher corresponding to the teacher to be assisted is determined from a famous teacher database, including: Determine the classroom support indicators and support intensity corresponding to the teacher to be supported; Using the support intensity, the support classroom indicators and a preset scoring system, determine the support expectation score of each famous teacher in the famous teacher database for the teacher; Based on the expected support score and the preset support learning curve effective interval, the famous teachers in the famous teacher database are matched and filtered to obtain a plurality of pre-recommended famous teachers; Determine the degree of compatibility between the teacher to be assisted and the multiple pre-recommended famous teachers, and based on the degree of compatibility, select a recommended famous teacher corresponding to the teacher from the multiple pre-recommended famous teachers.

10. The method according to claim 9, characterized in that The method of using the support intensity, the support classroom indicators and a preset scoring system to determine the support expectation score of each famous teacher in the famous teacher database for the teacher includes: According to the following formula 8, the support intensity, the support classroom index and the preset scoring system are used to determine the support expectation score of each famous teacher in the famous teacher database for the teacher: Formula 8 in, is the score of famous teacher M under the supporting classroom indicator i, is the score of teacher N under the support classroom indicator i; is the score of the teacher M's expectation of assistance to the teacher, k is the preset factor, is the support weight of supporting classroom indicator i.

Images