Intelligent English teaching method and system
By analyzing students' learning behaviors through big data and machine learning, an intelligent English teaching system is built, which solves the problems of individual differences and insufficient evaluation in traditional teaching methods, and realizes personalized teaching and efficient learning feedback.
Patent Information
- Application Number
- CN202510826132.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-19
- Publication Date
- 2025-09-26
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Traditional English teaching methods cannot meet individual differences, ignore the learning process and habit assessment, and lack data analysis tools, resulting in poor teaching results and failure to adapt to new learning methods brought about by technological development.
We use big data technology to analyze learning behavior data, use machine learning models to personalize course recommendations, combine speech recognition and natural language processing technologies for real-time feedback, and build an intelligent English teaching system, including learner management, data collection, course content, speech recognition, and evaluation modules.
It realizes personalized teaching, evaluates students' learning effects in multiple dimensions, improves teaching quality and learning efficiency, adjusts learning plans in real time, and provides personalized feedback and data-driven education strategies.
Smart Images
Figure CN120707350A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of intelligent teaching, and in particular relates to an intelligent English teaching method and system. Background Art
[0002] In today's globalized world, the importance of English as an international language has become increasingly prominent. However, traditional English teaching methods face many challenges and shortcomings in practice.
[0003] First, existing English instruction often adopts a one-size-fits-all approach, providing all students with the same course content and exercises without fully accounting for individual differences. This uniform approach struggles to meet the diverse needs and abilities of students, leading some to experience boredom and frustration, while others may struggle to develop effectively due to a lack of challenge. Especially in classes with widely varying abilities, teachers often struggle to accommodate each student's learning progress, resulting in poor learning outcomes for some.
[0004] Secondly, traditional assessment mechanisms primarily focus on testing language knowledge, often neglecting a comprehensive assessment of the learning process and habits. This single-dimensional approach not only fails to accurately reflect students' true learning status but can also lead to deviations in their learning strategies. For example, students may focus more on test scores and neglect improving their practical language application skills, which in turn affects the development of their comprehensive language application abilities.
[0005] Furthermore, the lack of effective data analysis tools in English teaching prevents teachers from obtaining timely feedback and information on students' learning progress. When designing and adjusting curriculum, teachers often rely on personal experience rather than analysis based on actual student learning data. This deficiency results in a lack of scientific basis for teaching decisions, which in turn affects the effectiveness of educational strategies.
[0006] Furthermore, with the rapid development of technology, many traditional teaching methods have not been updated to adapt to new learning styles. While some educational technologies have begun to be applied to English teaching, most systems still lack intelligent, personalized instruction. This has created an urgent need in the education community for methods and systems that can comprehensively leverage advanced technologies to enhance student learning experiences and outcomes. Summary of the Invention
[0007] In order to solve the above technical problems, the present invention provides an intelligent English teaching method and system to solve the problems raised in the background technology.
[0008] According to a first aspect of the present disclosure, an intelligent English teaching method is proposed, comprising the following steps:
[0009] S1. Use big data technology to collect and analyze students' learning behavior data, and use machine learning models to analyze students' English proficiency, learning habits, and interests;
[0010] S2. Based on the student analysis results, use collaborative filtering to recommend personalized courses and learning resources and dynamically adjust the learning path;
[0011] S3. Intelligently adjust course content based on students' learning progress and abilities, provide exercises and tests of varying difficulty levels, and introduce real-time updated English learning materials.
[0012] S4. Analyze classroom videos, provide coding schemes, automatically save teacher and student speech behaviors, and present behavior frequency distribution and transition timing in graphical form;
[0013] S5. Correct students’ pronunciation through speech recognition and synthesis technology, and automatically grade and provide feedback on students’ essays using natural language processing technology;
[0014] S6. Combine the decision tree algorithm and neural network to generate an English teaching evaluation model to comprehensively evaluate students' learning outcomes and teachers' teaching quality.
[0015] Preferably, the learning behavior data includes learning time, learning frequency, learning content and interaction status; the data collection sources include online course platforms and homework submitted by students.
[0016] Preferably, a supervised learning method is used to create a student profile, with the student's English proficiency, study habits, and interests as target variables, and features such as study time and study frequency extracted from existing student data as input variables. The supervised learning model is expressed as: y = Xβ + ε, where y represents the target variable, X represents the input variable, and ε represents the error term.
[0017] Clustering algorithms are used to group students. Students with similar learning behaviors are divided into the same cluster by calculating the distance or similarity between sample points. That is, by minimizing the sum of the squared distances from the sample points to their corresponding cluster centers, which can be expressed as: Where k represents the total number of clusters, C j represents the sample set contained in the jth cluster, x i represents the i-th sample point, μ j represents the center point of the jth cluster, ||x i -μ j || represents the sample point x i To the cluster center μ j Euclidean distance, analyze clustering results, and identify different learning behavior patterns according to the characteristics of different clusters;
[0018] A classification algorithm is used to analyze students' English proficiency, learning habits, and interests. The students' English proficiency, learning habits, and interests are used as target variables, and other features are used as input variables to train a classification model. The model is expressed as follows: Where P(y|X) represents the conditional probability of the target variable y given feature X, P(X|y) represents the probability distribution of feature X given the target variable y, P(y) represents the marginal probability distribution of the target variable y, and P(X) represents the marginal probability distribution of the target variable X. Through model prediction, a label or score is generated for each student, which includes their English proficiency, learning habits, and interests. The results of clustering and classification are used to generate a student profile.
[0019] Preferably, collaborative filtering calculates the similarity between students, finds other students similar to the target student, and predicts the courses and learning resources that the target student is interested in based on the preferences of similar students. The similarity between students is expressed as:
[0020] Among them, r xy represents the Pearson correlation coefficient between student x and student y, X i represents the score of student x in the i-th course, Y i represents the score of student y on the i-th course, n represents the number of scored courses, that is, the number of courses that students x and y have scored, represents the average score of student x in all courses, represents the average score of student y on all courses; then, based on the scores of similar students on a certain course, the target student’s score for the course is predicted, which is expressed as:
[0021]
[0022] Among them, P uc represents the predicted score of target student u for course c, sim(u,v) represents the similarity between target student u and similar student v, r uc represents the actual rating of similar student v on course c, and S represents the set of students similar to the target student u.
[0023] Preferably, the matching degree between students and courses is calculated based on the characteristics of students' English proficiency, learning habits and interests and the characteristics of courses or learning resources. Features including subject, difficulty and type are extracted for each course or learning resource. The course or learning resource is represented by a feature vector, and the similarity between it and the course is calculated as: Where c1 and c2 represent the eigenvectors of courses c1 and c2 respectively. Based on the students’ performance and learning progress, the learning path is dynamically adjusted to recommend the next learning content for the students.
[0024] Preferably, natural language generation technology is used to automatically generate exercises and tests based on students' learning progress and abilities, that is, questions of different difficulty levels are generated based on predefined templates and students' historical answer data; the predefined templates are established in advance through a library of exercise templates covering different grammar, vocabulary and language skills; and the latest English learning resources from the Internet are integrated to update the course content regularly.
[0025] Preferably, computer vision and speech recognition technology are used to analyze classroom recordings to automatically extract teacher-student interaction information, including students' speeches, questions, and teachers' responses. Coding standards are set to automatically mark teachers' and students' speeches, questions, and responses in class.
[0026] Computer vision uses target detection to identify students and teachers in the video, expressed as: in represents the set of predicted bounding boxes, f θ Represents the neural network model, I represents the input image; then the actions and postures of the student and teacher are recognized through posture estimation, which is expressed as: in represents the set of predicted key points, f φ Represents the neural network model, I represents the input image;
[0027] Speech recognition extracts MFCC features from audio signals and establishes an acoustic model to describe the relationship between speech signals and pronunciation units, which can be expressed as: Where P(O|λ) represents the likelihood of the predicted sequence under given model parameters, α i Represents the forward probability; at the same time, a language model is established to describe the probability distribution of word sequences, which is expressed as: in represents the probability of the i-th word appearing given the first n-1 words, Representing word sequences The number of times it appears in the corpus, Representing word sequences the number of times it occurs in the corpus;
[0028] By counting the number of times teachers and students speak, ask questions, and respond in class, a histogram or bar chart is drawn to represent the frequency distribution of behavior, and a time series graph of teacher-student interaction is drawn to represent the conversion sequence through the temporal changes in teacher-student interaction.
[0029] Preferably, a speech recognition API is used to monitor students’ pronunciation in real time and provide feedback. By converting students’ voice input into text and then comparing it with the standard text, pronunciation errors can be identified. The recognition algorithm is expressed as:
[0030]
[0031] Where O represents the student's speech signal, λ represents the parameters of the hidden Markov model, and q represents the hidden state sequence. Speech synthesis is then used to provide students with standard pronunciation demonstrations. The audio generation process is expressed as: y = f(x; θ), where y represents the synthesized speech signal, x represents the input text, and θ represents the parameters of the synthesis model.
[0032] The NLP model is used to automatically correct students' essays and provide feedback on grammar, spelling, and structure. The essays are analyzed and errors are identified. Word segmentation and part-of-speech tagging are used for text preprocessing. The BERT model is used to detect grammatical and spelling errors. This is expressed as follows:
[0033] P(y|x;θ)=softmax(W·E+b)
[0034] Among them, y represents the output label, which is wrong or correct, x represents the representation of the input text, W and b represent model parameters; and the correctly spelled word is predicted by the generative model. The prediction formula is expressed as:
[0035] y t =f(y t-1 ,z t ;θ)
[0036] Among them, y t Indicates the current word, z t Represents context information, and θ represents the model parameter; then through syntactic analysis, the sentence structure and paragraph logic are identified. The sentence structure is analyzed using a syntactic tree, which is expressed as:
[0037]
[0038] Among them, S represents a sentence, T represents a syntax tree, and w i Indicates vocabulary, parents(w i ) represents the parent node of the vocabulary; a feedback report is generated for the identified grammatical and spelling errors, structuring different types of errors and corresponding correction suggestions.
[0039] Preferably, an English teaching evaluation model combining a decision tree and a neural network is constructed. The decision tree is used to analyze students' learning data and determine key evaluation indicators. A data set of students' learning outcomes is constructed, including multiple indicators and student evaluation levels. Then, the information gain is used to select the optimal feature, which is expressed as: IG(D,A)=H(D)-H(D|A), where H(D) is the entropy of the data set D and H(D|A) represents the conditional entropy of the given feature A. By combining the key indicators selected by the decision tree, the neural network model is used to further analyze the students' learning outcomes. The output probability distribution represents the evaluation of the students' learning outcomes, which is expressed as: Where P(y i |x;θ) represents the probability that the model predicts that the student belongs to category i, z i Represents the linear output of one layer, and K represents the number of classifications. By analyzing the results of the model output, feedback on learning outcomes is provided to students, areas for improvement are pointed out, and data-driven strategies are provided to teachers.
[0040] According to a second aspect of the present disclosure, an intelligent English teaching system is proposed, comprising:
[0041] Learner management module, used to manage students' basic information and learning progress, and track and record each student's learning achievements, preferences, and progress;
[0042] Data collection and analysis module, used to collect students' learning data and feedback, and analyze learning behavior and results data;
[0043] Course content module, which provides English learning resources, including textbooks, exercises, videos and audio materials;
[0044] The speech recognition and assessment module receives and processes students' speech input, uses speech processing algorithms to analyze accuracy and fluency, and assesses students' spoken pronunciation and fluency in real time, providing feedback.
[0045] The homework correction module is used to automatically correct students' English compositions, analyze them, identify errors, and display error types and improvement directions through charts;
[0046] The evaluation module analyzes students' learning outcomes based on decision trees and neural networks and generates evaluation reports.
[0047] Compared with the prior art, the present invention has the following beneficial effects:
[0048] 1. By analyzing students' learning data, the system can provide each learner with personalized courses and exercises that meet their needs and abilities, and adjust the learning plan and content in real time based on the student's performance, helping students learn in an environment suitable for their level.
[0049] 2. This invention provides data-driven learning outcome feedback through an assessment model, helping students identify their strengths and weaknesses, thereby enabling more efficient and targeted learning. The multi-dimensional assessment not only assesses students' language proficiency but also focuses on their learning habits and methods, contributing to an all-round improvement in teaching quality.
[0050] 3. The present invention improves the intelligence of the system through technologies such as speech recognition, natural language processing and machine learning, making English learning more efficient and comprehensive, and can collect and analyze data in real time, and at the same time be used to improve educational strategies and teaching content. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] Figure 1 This is a flow chart of the intelligent English teaching method of the present invention;
[0052] Figure 2 This is a framework diagram of the intelligent English teaching system of the present invention. DETAILED DESCRIPTION
[0053] The following embodiments of the present invention are described in further detail with reference to the accompanying drawings and examples. The following examples are used to illustrate the present invention but are not intended to limit the scope of the present invention.
[0054] As attached Figure 1 As shown:
[0055] Embodiment 1: The present invention provides an intelligent English teaching method, comprising the following steps:
[0056] S1. Use big data technology to collect and analyze students' learning behavior data, and use machine learning models to analyze students' English proficiency, learning habits, and interests; learning behavior data includes learning time, learning frequency, learning content, and interaction; data collection sources include online course platforms and student-submitted assignments.
[0057] Use supervised learning to create student profiles. Take the student's English proficiency, study habits, and interests as target variables, and extract features such as study time and study frequency from existing student data as input variables. The supervised learning model is expressed as: y = Xβ + ε, where y represents the target variable, X represents the input variable, and ε represents the error term.
[0058] Clustering algorithms are used to group students. Students with similar learning behaviors are divided into the same cluster by calculating the distance or similarity between sample points. That is, by minimizing the sum of the squared distances from the sample points to their corresponding cluster centers, which can be expressed as: Where k represents the total number of clusters, C j represents the sample set contained in the jth cluster, x i represents the i-th sample point, μ jrepresents the center point of the jth cluster, ||x i -μ j || represents the sample point x i To the cluster center μ j Euclidean distance, analyze clustering results, and identify different learning behavior patterns according to the characteristics of different clusters;
[0059] A classification algorithm is used to analyze students' English proficiency, learning habits, and interests. The students' English proficiency, learning habits, and interests are used as target variables, and other features are used as input variables to train a classification model. The model is expressed as follows: Where P(y|X) represents the conditional probability of the target variable y given feature X, P(X|y) represents the probability distribution of feature X given the target variable y, P(y) represents the marginal probability distribution of the target variable y, and P(X) represents the marginal probability distribution of the target variable X. Through model prediction, a label or score is generated for each student, including their English proficiency, learning habits, and interests. Using the results of clustering and classification, a student profile is generated. This profile includes learning behavior characteristics, English proficiency levels, learning preferences, and other information, which can be used to facilitate subsequent personalized recommendations and teaching strategy adjustments.
[0060] S2. Based on the student analysis results, collaborative filtering is used to recommend personalized courses and learning resources, and dynamically adjust the learning path. Collaborative filtering is a technology that recommends personalized courses and learning resources based on user behavior data. Collaborative filtering calculates the similarity between students to find other students similar to the target student. Based on the preferences of similar students, the courses and learning resources that the target student would be interested in are predicted. The similarity between students is expressed as:
[0061]
[0062] Among them, r xy represents the Pearson correlation coefficient between student x and student y, X i represents the score of student x in the i-th course, Y i represents the score of student y on the i-th course, n represents the number of scored courses, that is, the number of courses that students x and y have scored, represents the average score of student x in all courses, represents the average score of student y on all courses; then, based on the scores of similar students on a certain course, the target student’s score for the course is predicted, which is expressed as:
[0063]
[0064] Among them, P ucrepresents the predicted score of target student u for course c, sim(u,v) represents the similarity between target student u and similar student v, r uc represents the actual rating of similar student v on course c, and S represents the set of students similar to the target student u.
[0065] S3. Intelligently adjust course content based on student learning progress and abilities, provide exercises and tests of varying difficulty levels, and introduce real-time updated English learning materials. Calculate the matching degree between students and courses based on the characteristics of students' English proficiency, learning habits, and interests, and the characteristics of courses or learning resources. Extract features for each course or learning resource, including topic, difficulty, and type. Use feature vectors to represent courses or learning resources. The similarity calculation between them and courses is expressed as: Where c1 and c2 represent the eigenvectors of courses c1 and c2, respectively. Based on student performance and learning progress, the learning path is dynamically adjusted to recommend the next step for the student. Dynamically adjusting the learning path monitors students' learning progress and performance in real time and dynamically adjusts the learning path based on their performance and learning progress.
[0066] Based on students' learning progress and abilities, natural language generation technology is used to automatically generate exercises and tests. This involves generating questions of varying difficulty levels based on predefined templates and students' historical answer data. The predefined templates are created by pre-establishing a library of exercise templates covering different grammar, vocabulary, and language skills. Course content is regularly updated by integrating the latest English learning resources from the internet, with a mechanism for automatically updating course content on a weekly or monthly basis. Collaborative filtering and content recommendation algorithms provide personalized course and learning resource recommendations for each student. Furthermore, combined with technology that dynamically adjusts learning paths, students' learning progress can be monitored in real time to ensure they receive the most appropriate learning content. This comprehensive approach effectively improves learning outcomes and student engagement. By leveraging natural language generation technology and a real-time update mechanism, course content can be intelligently adjusted to better meet students' individual learning needs. This approach not only enhances the learning experience but also stimulates student motivation, helping them achieve better learning outcomes in a shorter timeframe.
[0067] S4. Analyze classroom videos, provide coding schemes, automatically save teacher and student speech behaviors, and graphically present the behavior frequency distribution and transition sequence. Classroom video analysis combines computer vision, speech recognition, and data visualization technologies to automatically extract and analyze teacher-student interaction information. Specifically, computer vision and speech recognition technologies are used to analyze classroom recorded videos to automatically extract teacher-student interaction information, including students' speeches, questions, and teachers' responses. Coding standards are set to automatically mark teachers' and students' speeches, questions, and responses in class. Speeches are marked by voice recognition, questions are marked by detecting changes in the question's tone or specific keywords, and responses are marked by determining whether the speech content is related to others' speeches. At the same time, each mark is assigned a timestamp for time series analysis.
[0068] Computer vision uses target detection to identify students and teachers in the video, expressed as: in represents the set of predicted bounding boxes, f θ Represents the neural network model, I represents the input image; then the actions and postures of the student and teacher are recognized through posture estimation, which is expressed as: in represents the set of predicted key points, f φ Represents the neural network model, I represents the input image;
[0069] Speech recognition extracts MFCC features from audio signals and establishes an acoustic model to describe the relationship between speech signals and pronunciation units, which can be expressed as: Where P(O|λ) represents the likelihood of the predicted sequence under given model parameters, α i Represents the forward probability; at the same time, a language model is established to describe the probability distribution of word sequences, which is expressed as: in represents the probability of the i-th word appearing given the first n-1 words, Representing word sequences The number of times it appears in the corpus, Representing word sequences the number of times it occurs in the corpus;
[0070] By counting the number of times teachers and students speak, ask questions, and respond in class, a histogram or bar chart is plotted to represent the frequency distribution of these behaviors. A time series diagram of teacher-student interactions is then plotted, and the temporal changes in these interactions are used to represent the transitional timing. Classroom video analysis automatically extracts information about teacher-student interactions by integrating multiple technologies, including computer vision, speech recognition, and data visualization, enabling teachers to intuitively understand classroom dynamics. This approach not only improves the efficiency of classroom interaction analysis but also provides data support for educational decision-making.
[0071] S5. Correct students' pronunciation through speech recognition and synthesis technology, and automatically grade and provide feedback on students' essays using natural language processing technology. Use the speech recognition API to monitor students' pronunciation in real time and provide feedback. By converting students' voice input into text and comparing it with standard text, pronunciation errors can be identified. The recognition algorithm is expressed as follows:
[0072]
[0073] Where O represents the student's speech signal, λ represents the parameters of the hidden Markov model, and q represents the hidden state sequence. Speech synthesis is then used to provide students with standard pronunciation demonstrations. The audio generation process is expressed as: y = f(x; θ), where y represents the synthesized speech signal, x represents the input text, and θ represents the parameters of the synthesis model.
[0074] The NLP model is used to automatically correct students' essays and provide feedback on grammar, spelling, and structure. The essays are analyzed and errors are identified. Word segmentation and part-of-speech tagging are used for text preprocessing. The BERT model is used to detect grammatical and spelling errors. This is expressed as follows:
[0075] P(y|x;θ)=softmax(W·E+b)
[0076] Among them, y represents the output label, which is wrong or correct, x represents the representation of the input text, W and b represent model parameters; and the correctly spelled word is predicted by the generative model. The prediction formula is expressed as:
[0077] y t =f(y t-1 ,z t ;θ)
[0078] Among them, y t Indicates the current word, z t Represents context information, and θ represents the model parameter; then through syntactic analysis, the sentence structure and paragraph logic are identified. The sentence structure is analyzed using a syntactic tree, which is expressed as:
[0079]
[0080] Among them, S represents a sentence, T represents a syntax tree, and w i Indicates vocabulary, parents(w i) represents the parent node of a word; identified grammatical and spelling errors are generated into feedback reports, structuring different types of errors and corresponding correction suggestions. By integrating speech recognition and synthesis technologies with natural language processing, real-time monitoring and feedback on students' pronunciation and writing is provided, helping them not only to correct pronunciation errors promptly but also to improve their writing skills. This intelligent feedback mechanism facilitates personalized learning and timely tutoring, improving student learning outcomes.
[0081] S6. Combine the decision tree algorithm and neural network to generate an English teaching evaluation model to comprehensively evaluate students' learning outcomes and teachers' teaching quality;
[0082] Specifically, by constructing an English teaching evaluation model that combines decision trees and neural networks, the decision tree is used to analyze students' learning data and determine key evaluation indicators. A data set of students' learning outcomes is constructed, including multiple indicators and student evaluation levels. Then, the information gain is used to select the optimal feature, which is expressed as: IG(D,A)=H(D)-H(D|A), where H(D) is the entropy of the data set D and H(D|A) represents the conditional entropy of the given feature A. By combining the key indicators selected by the decision tree, the neural network model is used to further analyze the students' learning outcomes. The output probability distribution represents the evaluation of the students' learning outcomes, which is expressed as: Where P(y i |x;θ) represents the probability that the model predicts that the student belongs to category i, z i The model then analyzes the output of the model to provide students with feedback on their learning outcomes, identify areas for improvement, and inform teachers of data-driven strategies. These data-driven strategies include developing personalized teaching plans and adjusting teaching content. Specifically, this involves designing different teaching activities based on students' learning progress and focusing on areas of weakness to enhance student learning outcomes. This English teaching evaluation model, combining decision trees and neural networks, identifies key evaluation indicators, analyzes learning outcomes, and provides real-time feedback. This helps students and teachers effectively understand their learning performance and improve their learning strategies. This multi-layered evaluation mechanism ensures a more personalized and effective teaching experience.
[0083] As attached Figure 2 As shown:
[0084] Embodiment 2: The present invention further provides an intelligent English teaching system, which is applied in Embodiment 1 and includes:
[0085] Learner management module, used to manage students' basic information and learning progress, and track and record each student's learning achievements, preferences, and progress;
[0086] Data collection and analysis module, used to collect students' learning data and feedback, and analyze learning behavior and results data;
[0087] Course content module, which provides English learning resources, including textbooks, exercises, videos and audio materials;
[0088] The speech recognition and assessment module receives and processes students' speech input, uses speech processing algorithms to analyze accuracy and fluency, and assesses students' spoken pronunciation and fluency in real time, providing feedback.
[0089] The homework correction module is used to automatically correct students' English compositions, analyze them, identify errors, and display error types and improvement directions through charts;
[0090] The evaluation module analyzes students' learning outcomes based on decision trees and neural networks and generates evaluation reports.
[0091] The implementation process of the intelligent English teaching system corresponds one-to-one to the intelligent English teaching method of the above embodiment, and will not be described in detail in this embodiment.
[0092] It is important to note that the construction and arrangement of the present application as shown in a number of different exemplary embodiments are illustrative only. Although only a few embodiments are described in detail in this disclosure, those reading this disclosure will readily appreciate that many modifications are possible without materially departing from the novel teachings and advantages of the subject matter described in this application, and that other substitutions, modifications, changes, and omissions may be made in the design, operating conditions, and arrangement of the exemplary embodiments without departing from the scope of the present invention. Therefore, the present invention is not limited to specific embodiments, but extends to a variety of modifications that still fall within the scope of the appended claims.
[0093] Additionally, in order to provide a concise description of exemplary embodiments, all features of an actual embodiment (i.e., those features that are not relevant to the best mode presently contemplated for carrying out the invention or those that are not relevant to implementing the invention) may not be described.
[0094] It will be understood that in the development of any actual embodiment, as in any engineering or design project, numerous implementation-specific decisions may be made. Such a development effort may be complex and time-consuming, but for those of ordinary skill having the benefit of this disclosure, the development effort will be a routine task of design, fabrication, and production without undue experimentation.
[0095] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. An intelligent English teaching method, characterized in that: The following steps are involved: S1. Use big data technology to collect and analyze students' learning behavior data, and use machine learning models to analyze students' English proficiency, learning habits, and interests; S2. Based on the student analysis results, use collaborative filtering to recommend personalized courses and learning resources and dynamically adjust the learning path; S3. Intelligently adjust course content based on students' learning progress and abilities, provide exercises and tests of varying difficulty levels, and introduce real-time updated English learning materials. S4. Analyze classroom videos, provide coding schemes, automatically save teacher and student speech behaviors, and present behavior frequency distribution and transition timing in graphical form; S5. Correct students’ pronunciation through speech recognition and synthesis technology, and automatically grade and provide feedback on students’ essays using natural language processing technology; S6. Combine the decision tree algorithm and neural network to generate an English teaching evaluation model to comprehensively evaluate students' learning outcomes and teachers' teaching quality.
2. The intelligent English teaching method according to claim 1, wherein: The learning behavior data includes learning time, learning frequency, learning content and interaction status; the data collection sources include online course platforms and assignments submitted by students.
3. The intelligent English teaching method according to claim 1, wherein: Use supervised learning to create student profiles. Take the student's English proficiency, study habits, and interests as target variables, and extract features such as study time and study frequency from existing student data as input variables. The supervised learning model is expressed as: y = Xβ + ε, where y represents the target variable, X represents the input variable, and ε represents the error term. Clustering algorithms are used to group students. Students with similar learning behaviors are divided into the same cluster by calculating the distance or similarity between sample points. That is, by minimizing the sum of the squared distances from the sample points to their corresponding cluster centers, which can be expressed as: Where k represents the total number of clusters, C j represents the sample set contained in the jth cluster, x i represents the i-th sample point, μ j represents the center point of the jth cluster, ||x i -μ j || represents the sample point x i To the cluster center μ j Euclidean distance, analyze clustering results, and identify different learning behavior patterns according to the characteristics of different clusters; A classification algorithm is used to analyze students' English proficiency, learning habits, and interests. The students' English proficiency, learning habits, and interests are used as target variables, and other features are used as input variables to train a classification model. The model is expressed as follows: Where P(y|X) represents the conditional probability of the target variable y given feature X, P(X|y) represents the probability distribution of feature X given the target variable y, P(y) represents the marginal probability distribution of the target variable y, and P(X) represents the marginal probability distribution of the target variable X. Through model prediction, a label or score is generated for each student, which includes their English proficiency, learning habits, and interests. The results of clustering and classification are used to generate a student profile.
4. The intelligent English teaching method according to claim 1, wherein: Collaborative filtering calculates the similarity between students, finds other students similar to the target student, and predicts the courses and learning resources that the target student is interested in based on the preferences of similar students. The similarity between students is expressed as: Among them, r xy represents the Pearson correlation coefficient between student x and student y, X i represents the score of student x in the i-th course, Y i represents the score of student y on the i-th course, n represents the number of scored courses, that is, the number of courses that students x and y have scored, represents the average score of student x in all courses, represents the average score of student y on all courses; then, based on the scores of similar students on a certain course, the target student’s score for the course is predicted, which is expressed as: Among them, P uc represents the predicted score of target student u for course c, sim(u,v) represents the similarity between target student u and similar student v, r uc represents the actual rating of similar student v on course c, and S represents the set of students similar to the target student u.
5. The intelligent English teaching method according to claim 4, characterized in that: The matching degree between students and courses is calculated based on the characteristics of students' English proficiency, learning habits and interests, and the characteristics of courses or learning resources. Features including subject, difficulty and type are extracted for each course or learning resource. The course or learning resource is represented by a feature vector, and the similarity between it and the course is calculated as: Where c1 and c2 represent the eigenvectors of courses c1 and c2 respectively. Based on the students’ performance and learning progress, the learning path is dynamically adjusted to recommend the next learning content for the students.
6. The intelligent English teaching method according to claim 1, wherein: Based on students' learning progress and abilities, exercises and tests are automatically generated using natural language generation technology. This means questions of varying difficulty are generated based on predefined templates and students' historical answer data. The predefined templates are created by pre-establishing a library of practice templates covering different grammar, vocabulary and language skills; integrating the latest English learning resources from the Internet and updating the course content regularly.
7. The intelligent English teaching method according to claim 1, wherein: Use computer vision and speech recognition technology to analyze classroom recordings and automatically extract information about teacher-student interactions, including student speeches, questions, and teacher responses. Coding standards are set to automatically label teacher and student speeches, questions, and responses. Computer vision uses target detection to identify students and teachers in the video, expressed as: in represents the set of predicted bounding boxes, f θ Represents the neural network model, I represents the input image; then the actions and postures of the student and teacher are recognized through posture estimation, which is expressed as: in represents the set of predicted key points, f φ Represents the neural network model, I represents the input image; Speech recognition extracts MFCC features from audio signals and establishes an acoustic model to describe the relationship between speech signals and pronunciation units, which can be expressed as: Where P(O|λ) represents the likelihood of the predicted sequence under given model parameters, α i Represents the forward probability; at the same time, a language model is established to describe the probability distribution of word sequences, which is expressed as: in represents the probability of the i-th word appearing given the first n-1 words, Representing word sequences The number of times it appears in the corpus, Representing word sequences the number of times it occurs in the corpus; By counting the number of times teachers and students speak, ask questions, and respond in class, a histogram or bar chart is drawn to represent the frequency distribution of behavior, and a time series graph of teacher-student interaction is drawn to represent the conversion sequence through the temporal changes in teacher-student interaction.
8. The intelligent English teaching method according to claim 1, wherein: Using the speech recognition API, we monitor students’ pronunciation in real time and provide feedback. By converting students’ voice input into text and comparing it with the standard text, we can identify pronunciation errors. The recognition algorithm is expressed as follows: Where O represents the student's speech signal, λ represents the parameters of the hidden Markov model, and q represents the hidden state sequence. Speech synthesis is then used to provide students with standard pronunciation demonstrations. The audio generation process is expressed as: y = f(x; θ), where y represents the synthesized speech signal, x represents the input text, and θ represents the parameters of the synthesis model. The NLP model is used to automatically correct students' essays and provide feedback on grammar, spelling, and structure. The essays are analyzed and errors are identified. Word segmentation and part-of-speech tagging are used for text preprocessing. The BERT model is used to detect grammatical and spelling errors. This is expressed as follows: P(y|x;θ)=softmax(W·E+b) Among them, y represents the output label, which is wrong or correct, x represents the representation of the input text, W and b represent model parameters; and the correctly spelled word is predicted by the generative model. The prediction formula is expressed as: y t =f(y t-1 ,z t ;θ) Among them, y t Indicates the current word, z t Represents context information, and θ represents the model parameter; then through syntactic analysis, the sentence structure and paragraph logic are identified. The sentence structure is analyzed using a syntactic tree, which is expressed as: Among them, S represents a sentence, T represents a syntax tree, and w i Indicates vocabulary, parents(w i ) represents the parent node of the vocabulary; a feedback report is generated for the identified grammatical and spelling errors, structuring different types of errors and corresponding correction suggestions.
9. The intelligent English teaching method according to claim 1, wherein: An English teaching evaluation model combining decision trees and neural networks was constructed. Decision trees were used to analyze students' learning data and determine key evaluation indicators. A dataset of student learning outcomes was constructed, including multiple indicators and student evaluation levels. Information gain was then used to select the optimal features, expressed as: IG(D,A) = H(D) - H(D|A), where H(D) is the entropy of the dataset D and H(D|A) represents the conditional entropy of a given feature A. By combining the key indicators selected by the decision tree, a neural network model was used to further analyze students' learning outcomes. The output probability distribution represents the evaluation of students' learning outcomes, expressed as: Where P(y i |x; θ) represents the probability that the model predicts that the student belongs to category i, z i Represents the linear output of one layer, and K represents the number of classifications; By analyzing the output of the model, feedback on learning outcomes is provided to students, areas for improvement are pointed out, and data-driven strategies are provided to teachers.
10. An intelligent English teaching system, applied to the intelligent English teaching method according to any one of claims 1 to 9, characterized in that: include: Learner management module, used to manage students' basic information and learning progress, and track and record each student's learning achievements, preferences, and progress; Data collection and analysis module, used to collect students' learning data and feedback, and analyze learning behavior and results data; Course content module, which provides English learning resources, including textbooks, exercises, videos and audio materials; The speech recognition and assessment module receives and processes students' speech input, uses speech processing algorithms to analyze accuracy and fluency, and assesses students' spoken pronunciation and fluency in real time, providing feedback. The homework correction module is used to automatically correct students' English compositions, analyze them, identify errors, and display error types and improvement directions through charts; The evaluation module analyzes students' learning outcomes based on decision trees and neural networks and generates evaluation reports.