Teaching quality improvement method and device, equipment and storage medium

Abstract

The invention relates to a teaching quality improvement method and device, equipment and a storage medium. The method comprises: obtaining a teaching video; converting the teaching video into text information; classifying the text information by adopting a pre-trained classification model to obtain a classification result; performing statistical analysis and clustering analysis on the classification result to obtain knowledge performance characteristics of the teacher in teaching; and according to the knowledge performance characteristics, a knowledge improvement scheme is generated to improve the teaching quality of teachers. According to the method and the device, a natural language processing technology is utilized, multi-dimensional and multi-level measurement can be carried out on teaching behaviors, the comprehensiveness and the accuracy of detection are improved, and meanwhile, the detection time is also shortened; in addition, human interference is eliminated, and measurement objectivity and fairness are improved. And finally, a personalized knowledge improvement scheme is generated according to the teaching video, a professional growth path better conforming to personal characteristics of teachers can be planned for the teachers, and the teaching quality is effectively improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a teaching quality improvement method, device, equipment and storage medium. Background Art

[0002] TPACK is a theory about teacher knowledge proposed by Koehler and Mishra. The theory divides the knowledge of teachers in the information age into seven components. Generally speaking, there are three main measurement methods, namely, quantitative research methods based on questionnaire surveys, qualitative research methods based on content analysis, semi-structured interviews, and classroom observations, and mixed research methods of quantitative and qualitative research.

[0003] At present, the measurement and evaluation methods based on teachers' TPACK knowledge are usually based on traditional methods such as scales, interviews, and classroom observations. The measurement methods are relatively simple and have the disadvantages of limited measurement dimensions, strong subjectivity, and long time consumption, resulting in poor evaluation results. Summary of the invention

[0004] Based on this, it is necessary to provide a teaching quality improvement method, device, equipment and storage medium to address the above technical problems.

[0005] In a first aspect, the present application provides a method for improving teaching quality. The method comprises:

[0006] Get instructional videos;

[0007] Converting the teaching video into text information;

[0008] Using a pre-trained classification model to classify the text information to obtain a classification result;

[0009] Performing statistical analysis and cluster analysis on the classification results to obtain the knowledge performance characteristics of teachers in teaching;

[0010] Based on the knowledge representation characteristics, a knowledge improvement plan is generated to improve the teaching quality of teachers.

[0011] In one embodiment, the pre-training method of the classification model includes:

[0012] Constructing a training database and a knowledge database, and annotating the training database based on the knowledge database; wherein the knowledge database is constructed based on the TPACK framework;

[0013] Constructing the structure of the classification model; wherein the classification model includes a BERT model, an LSTM model, a multi-head self-attention model and a CRF model, and the LSTM model includes an input gate, a forget gate, a repeat gate and an output gate;

[0014] The classification model is iteratively trained according to the training database. In each training, the loss value of the output result of the classification model and the true result are compared, and the parameters of the classification model are gradually adjusted according to the loss value until the output result of the classification model meets the preset conditions, thereby obtaining a trained classification model.

[0015] In one embodiment, the pre-trained classification model classifies the text information, including:

[0016] The BERT model is used to encode the text information to obtain a representation vector for each word;

[0017] Using the LSTM model, the dependency relationship of the representation vector is obtained to obtain text features;

[0018] A multi-head self-attention model is used to obtain the relationship between the global information and local information of text features, and obtain text features with multi-angle information;

[0019] The CRF model is used to predict the probability distribution corresponding to the multi-angle text features to obtain the classification results.

[0020] In one embodiment, the knowledge database includes a keyword library; the method further includes:

[0021] Get alternative word library;

[0022] The Word2Vector model is used to calculate the similarity between the keyword library and the candidate word library, and words whose similarity meets the preset numerical requirements are selected from the candidate word library and added to the keyword library.

[0023] In one embodiment, the performing of statistical analysis and cluster analysis on the classification results to obtain the knowledge performance characteristics of the teacher in teaching includes:

[0024] Counting first data of different types of knowledge elements in the classification results, wherein the first data includes the type, frequency and duration of occurrence; the knowledge elements are obtained by parsing the TPACK framework, and the knowledge elements include any one or more of subject knowledge elements, teaching activity knowledge elements, teaching behavior knowledge elements, technical tool knowledge elements and technical tool use behavior knowledge elements;

[0025] Using a K-means clustering algorithm, cluster analysis is performed on the classification results to obtain second data, wherein the second data includes subcategories after the original categories are subdivided, and specific behavioral characteristics of the teacher under the original categories and the subcategories;

[0026] The first data and the second data constitute the knowledge representation feature.

[0027] In one embodiment, the method further comprises:

[0028] Based on the knowledge representation characteristics, a visual chart is generated for display.

[0029] In one embodiment, the method further comprises:

[0030] A question-answering engine is configured to match answers from a knowledge database according to the knowledge representation features and questions input by users, and present the answers to users.

[0031] In a second aspect, the present application also provides a teaching quality improvement device. The device comprises:

[0032] Acquisition module, used for teaching video; converting the teaching video into text information;

[0033] A classification module, used to classify the text information using a pre-trained classification model to obtain a classification result;

[0034] An analysis module, used to perform statistical analysis and cluster analysis on the classification results to obtain the knowledge performance characteristics of teachers in teaching;

[0035] The solution generation module is used to generate a knowledge improvement solution based on the knowledge representation characteristics to improve the teaching quality of teachers.

[0036] In a third aspect, the present application further provides a computer device, wherein the computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the method steps described in any one of the first aspects are implemented.

[0037] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method steps described in any one of the first aspects are implemented.

[0038] The above teaching quality improvement method, device, equipment and storage medium have at least the following advantages:

[0039] This application converts the acquired teaching videos into text information, uses a pre-trained class model to classify the text information, and obtains the classification results; then statistical analysis and cluster analysis are performed on the classification results to obtain the knowledge performance characteristics of teachers in teaching, and finally, based on the knowledge performance characteristics, a knowledge improvement plan is generated to improve the teaching quality of teachers. This application uses natural language processing technology to measure teaching behaviors in multiple dimensions and levels, which improves the comprehensiveness and accuracy of detection and also reduces detection time; in addition, eliminating human interference also improves the objectivity and fairness of measurement. Finally, generating a personalized knowledge improvement plan based on the teaching video can help teachers plan a professional growth path that is more in line with their personal characteristics, effectively improving the quality of teaching. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The drawings constituting a part of this application are used to provide a further understanding of the present invention. The illustrative embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0042] Figure 1 An application environment diagram of a teaching quality improvement method in an embodiment;

[0043] Figure 2 A flowchart of a method for improving teaching quality in one embodiment;

[0044] Figure 3 A schematic diagram of the composition of TPACK knowledge elements in an embodiment;

[0045] Figure 4 A structural diagram of an LSTM model that introduces a repetition gate in one embodiment;

[0046] Figure 5 is a structural block diagram of a teaching quality improvement device in one embodiment;

[0047] Figure 6 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0048] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0049] While some exemplary embodiments of the present invention have been described for the purpose of illustration, it should be understood that the present invention may be implemented in other ways not specifically shown in the drawings.

[0050] The teaching quality improvement method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown, the terminal 102 communicates with the server 104 through a network. The data storage system can store data that the server 104 needs to process. The data storage system can be integrated on the server 104, or placed on the cloud or other network servers.

[0051] Among them, the terminal 102 can send the recorded teaching video to the server 104, so that the server 104 processes the training sample, for example, the server 104 converts the teaching video into text information; uses the pre-trained classification model to classify the text information to obtain the classification result; performs statistical analysis and cluster analysis on the classification result to obtain the knowledge expression characteristics of the teacher in the teaching; generates a knowledge improvement plan based on the knowledge expression characteristics. In this way, the server 104 feeds back the knowledge improvement plan to the terminal 102 to improve the teaching quality of the teacher.

[0052] The above-mentioned teaching quality improvement method converts the acquired teaching video into text information, uses a pre-trained class model to classify the text information, and obtains the classification results; then statistical analysis and cluster analysis are performed on the classification results to obtain the knowledge performance characteristics of teachers in teaching, and finally, based on the knowledge performance characteristics, a knowledge improvement plan is generated to improve the teaching quality of teachers. This application uses natural language processing technology to measure teaching behaviors in multiple dimensions and levels, which improves the comprehensiveness and accuracy of detection and also reduces detection time; in addition, eliminating human interference also improves the objectivity and fairness of measurement. Finally, generating a personalized knowledge improvement plan based on the teaching video can plan a professional growth path for teachers that is more in line with their personal characteristics, effectively improving the quality of teaching.

[0053] The terminal 102 may be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, IoT devices, and portable wearable devices. The IoT devices may be smart speakers, smart TVs, smart air conditioners, smart car-mounted devices, etc. The portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers.

[0054] In an exemplary embodiment, the present application provides a method for improving teaching quality. Figure 1 The server 104 in FIG. 1 is used as an example for explanation.

[0055] See also Figure 2 , Figure 2 The following is a flow chart of a teaching quality improvement method according to the present embodiment, which specifically includes the following steps:

[0056] Step S202, obtaining a teaching video, and converting the teaching video into text information.

[0057] Specifically, the teaching video is recorded by the teacher during the teaching process. When recording, the voice in the video should be ensured to be clear. At the same time, when converting the teaching video into text information, the speaker should be marked to distinguish the speech of the teacher and the students.

[0058] Step S204: Use the pre-trained classification model to classify the text information to obtain a classification result.

[0059] Specifically, the classification model is trained based on the training database and the knowledge database. The training database comes from the text converted from the teaching videos publicly available on the Internet, or the text of the classroom teaching records. The knowledge database comes from text materials such as educational resources, teaching cases, technical tools, and TPACK-related research related to TPACK knowledge. After obtaining the text materials, the classroom teaching record text is preprocessed by natural language processing technology and named entity recognition tools, such as word segmentation and punctuation removal, and then the preprocessed text is entity labeled and text label classified to finally obtain the knowledge database. Furthermore, the knowledge database can also provide support for the subsequent development of the teacher TPACK knowledge intelligent question and answer system.

[0060] It should be noted that the TPACK framework in the embodiment of the present application includes three types of core knowledge, namely, subject knowledge (CK), pedagogical knowledge (PK) and technical knowledge (TK). These three types of knowledge are closely integrated with teaching elements such as teaching content, teaching activities, teaching activities, and technical tools in classroom teaching. By analyzing various teaching elements in classroom teaching, the teacher TPACK knowledge elements in classroom teaching are obtained, which can accurately measure and evaluate the teacher's TPACK level.

[0061] See also Figure 3 , Figure 3 The following is a schematic diagram of the composition of TPACK knowledge elements. Teachers' TPACK knowledge elements include any one or more of the subject content knowledge elements, teaching activity knowledge elements, teaching behavior knowledge elements, technology tool knowledge elements, and technology tool use behavior knowledge elements.

[0062] Among them, subject content is the main content of student learning and teacher teaching. In classroom teaching, subject content is the external manifestation of subject knowledge, and subject content knowledge elements are the teaching of relevant subject knowledge points in this lesson by teachers in classroom teaching. Subject content knowledge elements correspond to CK in the TPACK framework, which is the teaching of relevant subject knowledge points by teachers in classroom teaching. The CK level of teachers is analyzed and evaluated through the teaching of relevant subject knowledge points by teachers in classroom teaching.

[0063] Teaching activities in classroom teaching are purposeful and meaningful practical activities carried out in the classroom with the help of various media to achieve the expected teaching goals. Classroom teaching behavior is the way in which teachers can effectively complete teaching tasks and achieve educational goals in teaching activities. The teaching activity knowledge elements and teaching behavior knowledge elements correspond to the PK in the TPACK framework. The teaching activity knowledge element is the teacher's ability to reasonably arrange teaching activities in classroom teaching; the teaching behavior knowledge element is the teacher's ability to reasonably arrange teacher behavior, student behavior and teacher-student interaction behavior in classroom teaching; the teacher's PK level is analyzed and evaluated by the arrangement of teaching activities and teaching behaviors by the teacher in classroom teaching.

[0064] Technology tools are tools that provide technical support and services for teachers' teaching. Teaching resources that provide support for teachers' teaching also fall into the category of technology tools. The use of technology tools in classroom teaching is to better cooperate with the development of classroom teaching activities. The technology tool knowledge element and the technology tool use behavior knowledge element correspond to TK in the TPACK framework. The technology tool knowledge element is the ability of teachers to use technology tools in classroom teaching, and the technology tool use behavior knowledge element is the ability of teachers to use various technology tools reasonably in classroom teaching to improve teaching efficiency; the TK level of teachers can be analyzed and evaluated by analyzing the use of technology tools by teachers in classroom teaching.

[0065] Further, according to the knowledge database, the original text data in the training database is annotated, and the original text data and the annotated original text data constitute the training database. According to the preset ratio, the training database is divided into a training set, a validation set and a test set, wherein the training set is used for model training, the validation set is used to adjust hyperparameters and prevent overfitting, and the test set is used to evaluate the final performance of the model. The preset ratio in this embodiment is set to 6:2:2.

[0066] Furthermore, the present application also predetermines the classification dimensions of the classification model based on the knowledge database, so that the classification model can divide the input text into preset categories. For example, the classification standards for teaching activities are formulated based on Bloom's cognitive process dimensions and "Digital Bloom", and teaching activities are divided into six categories: memory, understanding, application, analysis, evaluation, and creation; through observation and analysis of the classroom teaching process, the use of teaching activities by teachers in classroom teaching is divided into three categories: expression display, teacher-student interaction, and teaching assistance.

[0067] The classification dimensions in the embodiments of the present application are divided into expression display, teacher-student interaction and teaching assistance.

[0068] Step S206, performing statistical analysis and cluster analysis on the classification results to obtain the knowledge performance characteristics of the teacher during teaching.

[0069] Specifically, the classification results of the above-mentioned classification model are based on the existing category labels to clearly classify the data. Since these category labels may not be sufficient to fully characterize the complex relationships within the data, this application further performs statistical analysis and cluster analysis on the text on the basis of the initial classification to determine whether there are fuzzy areas or overlapping characteristics between different categories, and whether there are similar but undefined new patterns or knowledge expression features in the text information, so as to make up for the shortcomings of the classification model.

[0070] Step S208, generating a knowledge improvement plan based on the knowledge representation characteristics to improve the teacher's teaching quality.

[0071] Specifically, the knowledge improvement plan is used to reflect the strengths and weaknesses of teachers in teaching, and to provide improvement suggestions for the weaknesses. For example, if a teacher's subject knowledge ratio is low according to the knowledge performance characteristics, it is recommended that the teacher increase the depth of knowledge explanation; if a teacher's one-way teaching time is long, it is recommended to increase the interactive session.

[0072] Furthermore, the embodiment of the present application also regularly collects teachers' implementation status and conducts quantitative analysis to evaluate the effectiveness of the improvement plan. Based on the analysis results, targeted feedback and suggestions are provided to teachers, and the improvement plan is adjusted to ensure its continuous improvement and optimization. The improvement plan includes learning resources, reflection strategies and other aspects to help teachers comprehensively improve their TPACK capabilities.

[0073] The above-mentioned teaching quality improvement method converts the acquired teaching video into text information, uses a pre-trained class model to classify the text information, and obtains the classification results; then performs statistical analysis and cluster analysis on the classification results to obtain the knowledge performance characteristics of teachers in teaching, and finally generates a knowledge improvement plan based on the knowledge performance characteristics to improve the teaching quality of teachers. This application combines the TPACK framework and uses natural language processing technology to measure teaching behaviors in multiple dimensions and levels, which improves the comprehensiveness and accuracy of detection and reduces detection time; in addition, eliminating human interference also improves the objectivity and fairness of measurement. Finally, generating a personalized knowledge improvement plan based on the teaching video can plan a professional growth path for teachers that is more in line with their personal characteristics, effectively improving the quality of teaching.

[0074] Optionally, the pre-training method of the classification model includes:

[0075] Build a training database and a knowledge database, and annotate the training database based on the knowledge database; the knowledge database is built based on the TPACK framework.

[0076] Construct the structure of the classification model; wherein the classification model includes the BERT model, the LSTM model, the multi-head self-attention model and the CRF model, and the LSTM model includes an input gate, a forget gate, a repeat gate and an output gate.

[0077] According to the training database, the classification model is iteratively trained. In each training, the loss value of the output result of the classification model and the true result is compared, and the parameters of the classification model are gradually adjusted according to the loss value until the output result of the classification model meets the preset conditions, thus obtaining a trained classification model.

[0078] Furthermore, the loss value of the above-mentioned comparison of the output result of the classification model and the true result is calculated according to the loss function, and the calculated loss value is transmitted back to each layer of the classification model through the back propagation algorithm to calculate the gradient of each layer; according to the calculated gradient, the optimizer is used to update the parameters of the model. Through continuous iterative training, the parameters of the model are gradually optimized and the loss value is gradually reduced until the output result of the classification model meets the preset conditions.

[0079] Optionally, the method for determining whether the preset condition is met includes: if the number of optimizations of the initial recognition model reaches a preset number, or the output accuracy of the initial recognition model meets a preset accuracy, then it is considered that the preset condition is met.

[0080] Furthermore, the classification models of the present application include a BERT model, an LSTM model with a repeated gate added, a CRF model, and a multi-head self-attention model, which are used to perform named entity recognition and text classification on the data set to obtain named entities and text sequences related to the teacher's TPACK knowledge.

[0081] Optionally, the pre-trained classification model classifies the text information, including:

[0082] The BERT model is used to encode the text information to obtain a representation vector for each word;

[0083] Using the LSTM model, the dependency relationship of the representation vector is obtained to obtain text features;

[0084] A multi-head self-attention model is used to obtain the relationship between the global information and local information of text features, and obtain text features with multi-angle information;

[0085] The CRF model is used to predict the probability distribution corresponding to the multi-angle text features to obtain the classification results.

[0086] Specifically, the BERT model encodes the input text information to obtain the representation vector of each word, and then inputs these vectors into LSTM for sequence modeling. The core of the BERT model lies in its encoder layer, which is composed of multiple bidirectional Transformer encoders stacked together. The bidirectional Transformer encoder can effectively capture the dependencies between contexts, thereby improving accuracy. Each encoder includes two sublayers, namely the multi-head self-attention layer and the feedforward neural network layer. The representation vector output by the BERT model is calculated based on mechanisms such as scaled point multiplication attention and multi-head self-attention. For each attention head, an independent linear transformation is used to obtain the query, key, and value matrices, and the self-attention of each attention head is calculated. The expression is:

[0087]

[0088] Where Q represents query, K represents key, and V represents value; k represents the key vector dimension;

[0089] The outputs of all attention heads are concatenated together and transformed through a linear layer, which is expressed as:

[0090] MultiHead(Q,K,V)=Concat(head1,...,headh)W O

[0091] Among them, the output expression of each attention head is:

[0092]

[0093] Among them, W i Q , W i K , W i V Respectively represent the learning weight matrix of the i-th head; W O Represents the final output layer weight matrix.

[0094] See also Figure 4 , Figure 4 The structure diagram of the LSTM model with the introduction of a repetition gate is shown, and the flow of information is controlled by introducing an input gate, a forget gate, and an output gate. Considering that the content of the classroom teaching text dataset in the present invention is the teacher's speech in classroom teaching, there are many repeated data, and the repeated data is often related to the corresponding teacher knowledge. These data are also important when analyzing the teacher's TPACK knowledge. Therefore, this study adds a repetition gate on the basis of the original gate mechanism of the LSTM model. The newly introduced repetition gate identifies the repeated elements in the input sequence through its internal calculation logic. The repetition pattern in the input sequence is captured by the learned weights and biases. In natural language processing tasks, especially those involving a large number of repeated elements, the introduction of a repetition gate helps to improve the performance of the model. By enhancing the impact of repeated elements on LSTM state updates, the model can focus more on processing repeated information, thereby more effectively capturing key features in the sequence.

[0095] Specifically, the forget gate is used to determine how much past information needs to be "forgotten" from the current memory cell state, and its expression is:

[0096] f_t=σ(W_f[x_t,h_{t-1}]+b_f)

[0097] Among them, f_t represents the forget gate at time t; σ represents the sigmoid function; W_f and b_f represent the weight and bias of the forget gate respectively; x_t represents the input vector at time t; h_{t-1} represents the hidden state vector at time t-1; the memory unit is used to store long-term information and is updated through the forget gate and the input gate.

[0098] The repetition gate is used to determine which important past information should be “repeated” in the current memory cell state. Its expression is:

[0099] r_t=σ(W_r[x_t,h_{t-1}]+b_r)

[0100] Among them, r_t represents the repeated gate at time t; W_r and b_r represent the weight and bias of the repeated gate respectively.

[0101] The input gate is used to determine how much new information needs to be "remembered" at the current time step and updated to the memory unit. Its expression is:

[0102] i_t=σ(W_i[x_t,h_{t-1}]+b_i)

[0103] Among them, i_t represents the input gate at time t; W_i and b_i represent the weight and bias of the input gate respectively.

[0104] The output of the input gate after repeated gate adjustment is:

[0105] i′_t=σ(i_t+alpha×r_t)

[0106] Among them, alpha is used to adjust the influence of the repeated gate output on the input gate output.

[0107] The updated state of the memory cell is:

[0108] Ct=ft×C_{t-1}+i_t×tanh(W_c[x_t,h_{t-1}]+b_c)

[0109] Among them, C_t represents the state of the memory cell at time t; C_{t-1} represents the state of the memory cell at time t-1; i′_t represents the output of the input gate after adjustment by the repeated gate; W_c and b_c represent the weight and bias of the cell state respectively; tanh represents the hyperbolic tangent function.

[0110] The output gate is used to determine which information of the current memory unit needs to be passed to the hidden state. Its expression is:

[0111] o_t=σ(W_o[x_t,h_{t-1]+b_o)

[0112] Among them, o_t represents the output gate at time t; W_o and b_o represent the weight and bias of the output gate respectively.

[0113] The hidden state is:

[0114] h_t=o_t×tanh(C_t)

[0115] Among them, h_t represents the hidden state at time t.

[0116] When the BERT model is combined with the LSTM model with a repeat gate, the representation vector of each word output by the BERT model is used as the input vector xt of the LSTM model. Then, the LSTM model performs sequence modeling through the above formula to obtain the hidden state ht of each time step. These hidden states can be used for subsequent named entity recognition and text classification tasks.

[0117] The multi-head self-attention model is an extended form of the attention mechanism. The model pays attention to any part of the input sequence and applies weights to it. The multi-head self-attention model uses multiple independent self-attention mechanisms to process text features in parallel. Each self-attention mechanism pays attention to a different part of the input sequence, thereby capturing the complex dependencies between different positions in the text. These mechanisms work together to simultaneously obtain global and local information of the text and generate text feature representations that contain multi-angle information.

[0118] The CRF model is a discriminative probabilistic undirected graph model that is used to predict the probability distribution of the corresponding output sequence for a given input sequence. In the CRF model, each node represents a state, and the edges between nodes represent the transition probability between states. The CRF model describes the relationship between the input sequence X and the output sequence Y by defining the conditional probability P(Y|X), where X represents the observation sequence and Y represents the state sequence. The relevant formula in the CRF model is:

[0119]

[0120] Where Z(X) represents the normalization factor;

[0121] i represents the position or index in the sequence; μ1 represents its weight; S1 represents the state characteristic function;

[0122] λ k represents the model parameter, which is the weight of the kth feature function;

[0123] T k (Y i-1 , Y i , X,i) represents the feature function, which is used to capture the relationship between the input sequence X and the output sequence Y.

[0124] k represents the number of characteristic functions.

[0125] In the above teaching quality improvement method, the BERT model encodes the input text information and extracts the semantic features of the text. The LSTM model can capture the sequence information and long-distance dependencies in the text information and extract the features of the text. The repetition gate in the LSTM model can enhance the processing effect of repeated data, more effectively capture the key features in the sequence, and improve the performance of the model. The multi-head self-attention model can focus on different parts of the text information from different angles, so that the model can better understand the relationship between the global information and local information of the text, and further improve the richness of the features. The CRF model can consider the transition probability between labels, so that the model is more in line with the actual semantic logic when predicting the label sequence. The classification model composed of the above models is used to classify the teaching text, and the teaching behavior can be measured in multiple dimensions and levels, which improves the comprehensiveness and accuracy of the detection.

[0126] Optionally, the knowledge database includes a keyword library. The teaching quality improvement method of the present application also includes:

[0127] Get alternative word libraries.

[0128] The Word2Vector model is used to calculate the similarity between the keyword library and the alternative word library, and words whose similarity meets the preset numerical requirements are selected from the alternative word library and added to the keyword library.

[0129] Specifically, in the process of training the classification model, this application also expands the knowledge database. By continuously expanding the database, it can comprehensively cover the different levels of expression of subject knowledge, providing strong support for subsequent text mining, teaching quality evaluation and teaching improvement.

[0130] Furthermore, the alternative vocabulary can come from a variety of sources such as curriculum standards, textbooks, teaching transcripts, and academic resources.

[0131] The keyword library is obtained by analyzing the curriculum standards and textbooks, combining the data in the classroom teaching transcript text, obtaining keywords and regular expressions of subject knowledge, and extracting words related to the TPACK framework from the text information. The keyword library is the core vocabulary of subject knowledge, and plays a key role in identifying whether the teacher has accurately and fully covered the key content in the curriculum standards and textbooks. In addition, the keyword library also provides a basis for text annotation and classification, helping the model to identify content fragments related to subject knowledge. It should be understood that different teacher TPACK knowledge elements include multiple different keywords. Exemplarily, for subject content knowledge elements, the keywords can be specific knowledge points such as "square root" or concept definitions such as "Pythagorean theorem".

[0132] Word2vector is used to calculate the similarity between word vectors for keyword expansion. In the Word2Vector model, the similarity between words is calculated by calculating the cosine similarity between two vectors. The formula for calculating cosine similarity is as follows:

[0133]

[0134] Among them, X(x1, x2, x3, ...xn), Y(y1, y2, y3, ...yn) are two vocabulary vectors. The greater the cosine similarity between the two vocabulary vectors, the higher the similarity between the two words. Select words whose similarity meets the preset numerical requirements and add them to the keyword library to complete the expansion of the keyword library.

[0135] Furthermore, after matching a suitable word, the frequency and type of the word are counted to determine the importance of the word.

[0136] The above-mentioned teaching quality improvement method expands the keyword library by calculating the similarity between the keyword library and the alternative word library, comprehensively covers the different levels of expression of subject knowledge, and provides strong support for subsequent text mining, teaching quality evaluation and teaching improvement.

[0137] Optionally, statistical analysis and cluster analysis are performed on the classification results to obtain the knowledge performance characteristics of teachers in teaching, including:

[0138] The first data of different types of knowledge elements in the classification results are counted, wherein the first data includes the type, frequency and duration of occurrence.

[0139] The K-means clustering algorithm is used to perform cluster analysis on the classification results to obtain second data, wherein the second data includes subcategories after the original categories are subdivided, and specific behavioral characteristics of the teacher under the original categories and subcategories.

[0140] The first data and the second data constitute the knowledge representation feature.

[0141] Specifically, the specific behavioral characteristics of teachers are a fine-grained portrayal of teachers' actual teaching behaviors in the classroom. These characteristics reflect the teachers' knowledge application methods and teaching styles in different situations. For example, for subject knowledge CK, the specific behavioral characteristics can be the definition or extension of core knowledge points, such as "functions are the corresponding relationship between variables" and "the changes in the function graph can help understand the supply and demand curves in economics." For teaching method knowledge PK, the specific behavioral characteristics can be organizing interactive discussions or providing feedback on students' answers, such as "five people form a group to discuss the derivation process of the function formula" and "this answer is close, but it lacks boundary conditions." For technical knowledge TK, the specific behavioral characteristics can be the use of tools for demonstration or teaching with the help of multimedia, such as "showing the changes in the function graph through PPT" and "playing animations to help students understand atomic structure."

[0142] Furthermore, the expression of the K-means clustering algorithm is:

[0143]

[0144] Among them, the value of W represents the quality of the clustering result;

[0145] K is the number of clusters; x is the data object;

[0146] C i is the i-th cluster center;

[0147] n is the dimension of the data object;

[0148] x j , C ij is x and C i The j-th attribute value of .

[0149] The above teaching quality improvement method statistically analyzes the first data obtained to characterize the type, frequency and duration of occurrence of different types of knowledge elements. Based on the first data, the teacher's time distribution on different types can be statistically analyzed to determine whether the teacher is overly dependent on a certain type of knowledge. In addition, statistical frequency distribution of different types can analyze the logic and coherence of the teaching process and determine whether the teacher's teaching arrangement is reasonable.

[0150] The second data obtained by the clustering algorithm represents the subcategories after the original categories are subdivided, as well as the specific behavioral characteristics of the teacher under the original categories and subcategories. By analyzing the specific behavioral characteristics, we can judge the teacher's teaching style, such as whether it focuses on conceptual explanation or practical application, whether it focuses on interaction with students or guiding students to learn independently; by analyzing the specific behavioral characteristics, we can also find the shortcomings of the teacher's behavioral characteristics under a certain type, such as insufficient use of technical tools, or long explanation time and less interaction.

[0151] By analyzing the above-mentioned knowledge representation characteristics, we can discover the strengths or weaknesses of teachers in the teaching process, thereby generating personalized knowledge improvement plans to help teachers improve the quality of teaching.

[0152] Optionally, the teaching quality improvement method of the present application further includes:

[0153] Generate visual charts for display based on knowledge representation characteristics.

[0154] Specifically, visualization charts can have various forms. For example, pie charts can be used to display the percentage distribution of CK, PK, and TK; bar charts can be used to display the frequency of different categories in different teaching stages; heat maps can be used to display the combination frequency between different categories; and radar charts can be used to display the performance intensity of teachers in the three dimensions of CK, PK, and TK.

[0155] The above teaching quality improvement method realizes the visualization of subject knowledge, teaching method knowledge and technical knowledge through the above visualization charts. Teachers can intuitively feel the data distribution and quickly assist in teaching quality evaluation and improvement.

[0156] Optionally, the teaching quality improvement method of the present application further includes:

[0157] Configure the question-answering engine to match answers from the knowledge database based on knowledge representation features and questions entered by users, and display them to users.

[0158] Specifically, the question and answer engine of the embodiment of the present application is built based on the NLP and question and answer engine services provided by the Baidu Smart Cloud Qianfan platform.

[0159] It should be understood that the knowledge database includes a keyword library, and each word in the keyword library is related to the three core knowledge (CK, PK, TK) of TPACK and their combinations; further, the knowledge database also includes text data, such as curriculum standards and textbook content, teaching cases, instructions for use of technical tools, etc.; further, the knowledge database also includes a variety of resource data related to teaching content, such as courseware and multimedia resources, homework and exercises, teaching evaluation tools, etc. Each keyword or text constitutes multiple nodes of the knowledge database. By constructing a structured knowledge graph of nodes and edges, the semantic relationship between knowledge can be displayed. When receiving the user's query statement, the keywords in the sentence are identified to match the nodes in the knowledge graph. If the keyword directly corresponds to a node in the knowledge graph, the information of the node is directly queried; if the keyword involves multiple nodes or combined knowledge, the relationship between the nodes is queried, so as to obtain the query results and feedback to the user.

[0160] Furthermore, the embodiment of the present application uses Python to create a graphical user interface to implement the design of the user login interface, and designs the account and password input boxes and the login button in the interface. By calling the API interface of Baidu Smart Cloud Qianfan, the user identity authentication and login functions are implemented.

[0161] Furthermore, the embodiment of the present application continuously updates and optimizes the knowledge database to improve the accuracy and coverage of the intelligent question and answer system.

[0162] The above-mentioned teaching quality improvement method converts the acquired teaching video into text information, uses a pre-trained class model to classify the text information, and obtains the classification results; then performs statistical analysis and cluster analysis on the classification results to obtain the knowledge performance characteristics of teachers in teaching, and finally generates a knowledge improvement plan based on the knowledge performance characteristics to improve the teaching quality of teachers. This application combines the TPACK framework and uses natural language processing technology to measure teaching behaviors in multiple dimensions and levels, which improves the comprehensiveness and accuracy of detection and reduces detection time; in addition, eliminating human interference also improves the objectivity and fairness of measurement. Finally, generating a personalized knowledge improvement plan based on the teaching video can plan a professional growth path for teachers that is more in line with their personal characteristics, effectively improving the quality of teaching.

[0163] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0164] Based on the same inventive concept, the embodiment of the present application also provides a teaching quality improvement device for implementing the teaching quality improvement method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more teaching quality improvement device embodiments provided below can refer to the limitations of the teaching quality improvement method above, and will not be repeated here.

[0165] See also Figure 5 In an exemplary embodiment, the present application provides a teaching quality improvement device, including: an acquisition module, a classification module, an analysis module and a solution generation module, wherein:

[0166] Acquisition module, used for teaching videos; converting teaching videos into text information;

[0167] The classification module is used to classify text information using a pre-trained classification model to obtain classification results;

[0168] The analysis module is used to perform statistical analysis and cluster analysis on the classification results to obtain the knowledge performance characteristics of teachers in teaching;

[0169] The solution generation module is used to generate knowledge improvement solutions based on knowledge representation characteristics to improve teachers' teaching quality.

[0170] Optionally, the teaching quality improvement device of the embodiment of the present application further includes: a training module.

[0171] The training module is used to pre-train the classification model, and the pre-training steps include: constructing a training database and a knowledge database, and annotating the training database based on the knowledge database; wherein the knowledge database is constructed based on the TPACK framework. Constructing the structure of the classification model; wherein the classification model includes a BERT model, an LSTM model, a multi-head self-attention model, and a CRF model, and the LSTM model includes an input gate, a forget gate, a repeat gate, and an output gate. According to the training database, the classification model is iteratively trained. In each training, the loss value of the output result of the classification model and the true result is compared, and the parameters of the classification model are gradually adjusted according to the loss value until the output result of the classification model meets the preset conditions, and a trained classification model is obtained.

[0172] Optionally, the method for determining whether the preset condition is met includes: if the number of optimizations of the initial recognition model reaches a preset number, or the output accuracy of the initial recognition model meets a preset accuracy, then it is considered that the preset condition is met.

[0173] Optionally, the classification module uses a pre-trained classification model to classify the text information, including: using a BERT model to encode the text information to obtain a representation vector for each word; using an LSTM model to obtain the dependency relationship of the representation vector to obtain text features; using a multi-head self-attention model to obtain the relationship between the global information and local information of the text features to obtain text features of multi-angle information; using a CRF model to predict the probability distribution corresponding to the multi-angle text features to obtain a classification result.

[0174] Optionally, the training module is also used to obtain an alternative vocabulary library; using the Word2Vector model, the similarity between the keyword library and the alternative vocabulary library is calculated, and words whose similarity meets preset numerical requirements are selected from the alternative vocabulary library and added to the keyword library.

[0175] Optionally, after matching a suitable word, the training module also performs frequency and type statistics on the word to determine the importance of the word.

[0176] Optionally, the analysis module performs statistical analysis and cluster analysis on the classification results to obtain the knowledge expression characteristics of the teacher in teaching, including: counting the first data of different types of knowledge elements in the classification results, wherein the first data includes the type, frequency and duration of occurrence; using the K-means clustering algorithm to perform cluster analysis on the classification results to obtain second data, wherein the second data includes sub-categories after the original categories are subdivided, and the specific behavioral characteristics of the teacher under the original categories and sub-categories; the first data and the second data constitute the knowledge expression characteristics.

[0177] Optionally, the teaching quality improvement device of the embodiment of the present application further includes: a display module.

[0178] The display module is used to generate visual charts for display based on the knowledge representation characteristics.

[0179] Optionally, the teaching quality improvement device of the embodiment of the present application further includes: a question and answer module.

[0180] The question-answering module is used to match answers from the knowledge database based on knowledge representation features and questions input by users, and display them to users.

[0181] The above-mentioned teaching quality improvement device converts the acquired teaching video into text information, uses a pre-trained class model to classify the text information, and obtains the classification results; then performs statistical analysis and cluster analysis on the classification results to obtain the knowledge performance characteristics of teachers in teaching, and finally generates a knowledge improvement plan based on the knowledge performance characteristics to improve the teaching quality of teachers. This application combines the TPACK framework and uses natural language processing technology to measure teaching behaviors in multiple dimensions and levels, which improves the comprehensiveness and accuracy of detection and also reduces detection time; in addition, eliminating human interference also improves the objectivity and fairness of measurement. Finally, generating a personalized knowledge improvement plan based on the teaching video can plan a professional growth path for teachers that is more in line with their personal characteristics, effectively improving the quality of teaching.

[0182] Each module in the above teaching quality improvement device can be implemented in whole or in part by software, hardware and their combination. Each module can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to each module.

[0183] Optionally, in an exemplary embodiment, the present application embodiment provides an intelligent question-answering system for generating corresponding query results according to the user's query statement and feeding back to the user. Among them, building the intelligent question-answering system includes the following steps:

[0184] Step 1: Build the login interface. Use Python to create a graphical user interface to implement the design of the user login interface. Design the account and password input boxes and the login button in the interface. Implement user identity authentication and login functions by calling the API interface of Baidu Smart Cloud Qianfan.

[0185] Step 2: Add data sets and structured processing, structure the teacher TPACK knowledge data set, including data cleaning, labeling, format conversion, etc. Integrate the processed data set into the knowledge database of the intelligent question-answering system, and continuously update and optimize the data set to improve the accuracy and coverage of the intelligent question-answering system.

[0186] Step 3: Integrate the intelligent question-answering system. Use the NLP and question-answering engine services provided by Baidu Smart Cloud Qianfan Platform to build an intelligent question-answering system. Import the structured knowledge base formed by the teacher's TPACK knowledge data set, configure the question-answering engine, match and infer in the knowledge base based on the questions entered by the user, and give the most appropriate answers.

[0187] Step 4: Write prompt words. According to the analysis results of the teacher's TPACK knowledge dashboard, write corresponding prompt words and embed the prompt words into the question-answering system to provide users with more accurate and valuable feedback and suggestions. Continuously update and optimize the prompt words to ensure their accuracy and practicality.

[0188] Step 5: Personalized content recommendation. Allow users to upload teaching videos and use the above-mentioned teaching quality improvement method to generate analysis results based on the teaching videos. After the user logs in, the analysis results of the teaching video are obtained according to the user login information. Based on the analysis results, a content-based text recommendation algorithm is used to generate personalized recommended content for the user. Text recommendation extracts features and calculates similarity of text content to find the content that best matches the teacher's needs. The main steps include content representation, feature learning, and generating a recommendation list.

[0189] Step 6: Provide teachers with TPACK knowledge improvement plans. Based on the intelligent question-and-answer system's answers to user questions and the analysis results of teaching videos, provide users with targeted TPACK knowledge improvement plans and suggestions. The plans include learning resources, reflection strategies and other aspects to help teachers comprehensively improve their teaching abilities.

[0190] Step 7: Continuously update and optimize, collect user feedback and opinions, evaluate and improve the performance and user experience of the intelligent question-answering system, continuously update and optimize the data set, prompt words and personalized recommendation algorithms, and regularly maintain and upgrade the intelligent question-answering system to ensure its stability and reliability.

[0191] The intelligent question-answering system can match and infer the questions in the knowledge base according to the questions input by the user and give the most appropriate answers. It can also provide users with targeted TPACK knowledge improvement plans and suggestions, including learning resources, reflection strategies and other aspects, to help teachers improve their teaching ability in an all-round way.

[0192] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 6 As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a method for improving teaching quality is implemented. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0193] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0194] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned teaching quality improvement method when executing the computer program.

[0195] In an exemplary embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned teaching quality improvement method are implemented.

[0196] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0197] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0198] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for improving teaching quality, characterized in that: The method comprises: Get instructional videos; Converting the teaching video into text information; Using a pre-trained classification model to classify the text information to obtain a classification result; Performing statistical analysis and cluster analysis on the classification results to obtain the knowledge performance characteristics of teachers in teaching; Based on the knowledge representation characteristics, a knowledge improvement plan is generated to improve the teaching quality of teachers.

2. The method according to claim 1, characterized in that The pre-training method of the classification model includes: Constructing a training database and a knowledge database, and annotating the training database based on the knowledge database; wherein the knowledge database is constructed based on the TPACK framework; Constructing the structure of the classification model; wherein the classification model includes a BERT model, an LSTM model, a multi-head self-attention model and a CRF model, and the LSTM model includes an input gate, a forget gate, a repeat gate and an output gate; The classification model is iteratively trained according to the training database. In each training, the loss value of the output result of the classification model and the true result are compared, and the parameters of the classification model are gradually adjusted according to the loss value until the output result of the classification model meets the preset conditions, thereby obtaining a trained classification model.

3. The method according to claim 2, characterized in that The pre-trained classification model classifies the text information, including: The BERT model is used to encode the text information to obtain a representation vector for each word; Using the LSTM model, the dependency relationship of the representation vector is obtained to obtain text features; A multi-head self-attention model is used to obtain the relationship between the global information and local information of text features, and obtain text features with multi-angle information; The CRF model is used to predict the probability distribution corresponding to the multi-angle text features to obtain the classification results.

4. The method according to claim 2, characterized in that: The knowledge database includes a keyword library; the method further includes: Get alternative word library; The Word2Vector model is used to calculate the similarity between the keyword library and the candidate word library, and words whose similarity meets the preset numerical requirements are selected from the candidate word library and added to the keyword library.

5. The method according to claim 1, characterized in that The statistical analysis and cluster analysis of the classification results are performed to obtain the knowledge performance characteristics of the teacher in the teaching, including: Counting first data of different types of knowledge elements in the classification results, wherein the first data includes the type, frequency and duration of occurrence; the knowledge elements are obtained by parsing the TPACK framework, and the knowledge elements include any one or more of subject knowledge elements, teaching activity knowledge elements, teaching behavior knowledge elements, technical tool knowledge elements and technical tool use behavior knowledge elements; Using a K-means clustering algorithm, cluster analysis is performed on the classification results to obtain second data, wherein the second data includes subcategories after the original categories are subdivided, and specific behavioral characteristics of the teacher under the original categories and the subcategories; The first data and the second data constitute the knowledge representation feature.

6. The method according to claim 5, characterized in that The method further comprises: Based on the knowledge representation characteristics, a visual chart is generated for display.

7. The method according to claim 1, characterized in that The method further comprises: A question-answering engine is configured to match answers from a knowledge database according to the knowledge representation features and questions input by users, and present the answers to users.

8. A teaching quality improvement device, characterized in that: The device comprises: Acquisition module, used for teaching video; converting the teaching video into text information; A classification module, used to classify the text information using a pre-trained classification model to obtain a classification result; An analysis module, used to perform statistical analysis and cluster analysis on the classification results to obtain the knowledge performance characteristics of teachers in teaching; The solution generation module is used to generate a knowledge improvement solution based on the knowledge representation characteristics to improve the teaching quality of teachers.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images