An English teaching management method and system based on semantic understanding and cloud services

Abstract

The present invention provides an English teaching management method and system based on semantic understanding and cloud services, which relate to the field of electronic data processing. The method comprises: establishing a cloud-based teaching resource library; determining multiple student groups based on learning pattern tendency data, English proficiency level data, and a weighted graph model of points of interest of multiple students, and generating basic teaching plans corresponding to the student groups; retrieving teaching resources from the cloud-based teaching resource library based on the basic teaching plans corresponding to the student groups, and pushing the corresponding teaching resources to the student terminals corresponding to the student groups according to the teaching progress of the basic teaching plans of the student groups; generating supplementary push content corresponding to the student groups based on real-time relevant information and the learning pattern tendency data, English proficiency level data, and a weighted graph model of points of interest of multiple students included in the student groups using semantic understanding technology, and pushing the supplementary push content to the student terminals corresponding to the student groups, thereby improving the efficiency and quality of English teaching.

Description

Technical Field

[0001] The present invention relates to the field of electronic data processing, and in particular to an English teaching management method and system based on semantic understanding and cloud services. Background Art

[0002] English is a universal language. As one of the most widely used official languages ​​in the world, it is the most advantageous tool for communication with other countries. Therefore, with the development of society and economic progress, people's demand for learning English is increasing.

[0003] Existing English teaching methods often rely on traditional manual instruction through textbooks. This approach places a high demand on English teachers and requires high-quality teachers. Limited by venue, it hinders widespread adoption and expansion. Furthermore, existing English teaching management practices may focus on a single aspect of English learning, such as grammar or vocabulary, lacking flexibility and personalization. These approaches are unable to adapt to students' learning progress and individual needs, and lack comprehensive support for English learning, resulting in low efficiency and quality of English teaching.

[0004] Therefore, it is necessary to provide an English teaching management method and system based on semantic understanding and cloud services to improve the efficiency and quality of English teaching. Summary of the Invention

[0005] The present invention provides an English teaching management method based on semantic understanding and cloud services, comprising: establishing a cloud-based teaching resource library; obtaining learning pattern tendency data, English proficiency level data, and a point-of-interest weighted graph model for a plurality of students; grouping the plurality of students according to the learning pattern tendency data, English proficiency level data, and the point-of-interest weighted graph model to determine a plurality of student groups; generating, for each student group, a basic teaching plan corresponding to the student group according to the learning pattern tendency data, English proficiency level data, and the point-of-interest weighted graph model of the plurality of students included in the student group; retrieving teaching resources from the cloud-based teaching resource library according to the basic teaching plan corresponding to the student group for each student group, and pushing the corresponding teaching resources to a student terminal corresponding to the student group according to the teaching progress of the basic teaching plan of the student group; obtaining real-time relevant information, wherein the real-time relevant information includes at least campus activity information and / or real-time news; and generating, for each student group, supplementary push content corresponding to the student group using semantic understanding technology based on the real-time relevant information and the learning pattern tendency data, English proficiency level data, and the point-of-interest weighted graph model of the plurality of students included in the student group, and pushing the supplementary push content to the student terminal corresponding to the student group.

[0006] Furthermore, learning pattern tendency data, English proficiency level data and interest point weight graph models of multiple students are obtained, including: obtaining English learning test data of multiple students under multiple teaching methods, and generating students' learning pattern tendency data based on the students' English learning test data; obtaining vocabulary test data, reading test data, oral test data and writing test data of multiple students, and generating students' English proficiency level data based on the students' vocabulary test data, reading test data, oral test data and writing test data; establishing interest point weight graph models for multiple sample students; obtaining interest point test data of multiple students, and establishing students' interest point weight graph models based on the students' interest point test data and the interest point weight graph models of multiple sample students.

[0007] Furthermore, an interest point weighted map model of multiple sample students is established, including: obtaining interest point test data of multiple test students; dividing the multiple test students into multiple test student groups according to the interest point test data of the multiple test students; for each test student group, establishing an interest point weighted map model of the sample students corresponding to the test student group based on the interest point test data of the multiple test students included in the test student group; establishing an interest point weighted map model of the students according to the students' interest point test data and the interest point weighted map models of the multiple sample students, including: establishing an initial interest point weighted map model of the students according to the students' interest point test data; determining similar test student groups according to the students' interest point test data and the interest point test data of the multiple test students included in each test student group; completing the initial interest point weighted map model of the students based on the interest point weighted map models of the sample students corresponding to the similar test student groups, and establishing the interest point weighted map model of the students.

[0008] Furthermore, the student's oral test data includes oral pronunciation test audio and oral dialogue test audio; based on the student's vocabulary test data, reading test data, oral test data and writing test data, generating the student's English proficiency level data, including: performing variational modal decomposition on the student's oral pronunciation test audio to extract the student's key sound features and pronunciation key features; based on the student's key sound features, obtaining similar standard oral pronunciation audio from a standard oral pronunciation audio library; performing variational modal decomposition on the similar standard oral pronunciation audio to extract standard pronunciation key features; determining the student's oral pronunciation level based on the student's pronunciation key features and standard pronunciation key features; determining the student's oral vocabulary features based on the student's oral dialogue test audio through a vocabulary recognition model; determining the student's oral grammar features based on the student's oral dialogue test audio through a grammar recognition model; determining the student's conversation emotion features based on the student's oral dialogue test audio through an emotion recognition model; determining the student's oral dialogue level based on the student's oral vocabulary features, oral grammar features and conversation emotion features, wherein the student's English proficiency level data at least includes the student's oral pronunciation level and oral dialogue level.

[0009] Furthermore, based on the learning pattern tendency data, English proficiency level data and interest point weight map model of multiple students, the multiple students are grouped to determine multiple student groups, including: for any two students, according to the interest point weight map model of the two students, calculating the similarity of the interest point weight map model of the two students; based on the similarity of the interest point weight map model of any two students, grouping the multiple students for the first time to determine multiple first student groups; for each first student group, according to the English proficiency level data of any two students included in the first student group, calculating the similarity of the English proficiency levels of any two students, and based on the similarity of the English proficiency levels of any two students, grouping the multiple students included in the first student group for the second time to determine multiple second student groups; for each second student group, according to the learning pattern tendency data of any two students included in the second student group, calculating the similarity of the learning pattern tendency of any two students, and based on the similarity of the learning pattern tendency of any two students, grouping the multiple students included in the second student group for the third time to determine multiple student groups.

[0010] Furthermore, based on the learning pattern tendency data, English proficiency level data and interest point weight graph model of the multiple students included in the student group, a basic teaching plan corresponding to the student group is generated, including: determining the target teaching method corresponding to the student group based on the learning pattern tendency data of the multiple student terminals included in the student group; determining the multiple target English knowledge points corresponding to the student group based on the English proficiency level data of the multiple student terminals included in the student group; generating the basic teaching plan corresponding to the student group based on the interest point weight graph model and multiple target English knowledge points of the multiple students included in the student group, wherein the basic teaching plan at least includes the target English knowledge points, target teaching methods and teaching objectives of multiple teaching time nodes.

[0011] Furthermore, obtaining real-time related information includes: determining a first target website, and obtaining candidate campus activity information from the first target website based on crawler technology; determining a second target website, and obtaining candidate real-time news from the second target website based on crawler technology; screening the candidate campus activity information and candidate real-time news to determine valid campus activity information and valid real-time news.

[0012] Furthermore, the candidate campus activity information and candidate real-time news are screened to determine valid campus activity information and valid real-time news, including: for each student group, generating a group interest point weight map model corresponding to the student group based on a plurality of students included in the student group; determining the interest points corresponding to the candidate campus activity information; determining the interest points corresponding to the candidate real-time news; screening the candidate campus activity information and candidate real-time news based on the group interest point weight map model corresponding to each student group, the interest points corresponding to the candidate campus activity information, and the interest points corresponding to the candidate real-time news to determine valid campus activity information and valid real-time news.

[0013] Furthermore, based on real-time relevant information and the learning pattern tendency data, English proficiency level data and interest point weight graph model of the multiple students included in the student group, semantic understanding technology is used to generate supplementary push content corresponding to the student group, and the supplementary push content is pushed to the student terminal corresponding to the student group, including: for each student group, based on the interest points corresponding to the valid campus activity information or the interest points corresponding to the valid real-time news, the target campus activity information or target real-time news is determined, and English reading articles are generated according to the English proficiency level data of the multiple students included in the student group and the target campus activity information or target real-time news using semantic understanding technology, and the English reading articles are converted into supplementary push content based on the learning pattern tendency data of the multiple students included in the student group, and the supplementary push content is pushed to the student terminal corresponding to the student group.

[0014] The present invention provides an English teaching management system based on semantic understanding and cloud services, which is applied to the above-mentioned English teaching management method based on semantic understanding and cloud services, including: a resource management module for establishing a cloud-based teaching resource library; a teaching management module for obtaining learning pattern tendency data, English proficiency level data and interest point weighted graph models of multiple students; grouping the multiple students according to the learning pattern tendency data, English proficiency level data and interest point weighted graph models of the multiple students to determine multiple student groups; for each student group, generating a basic teaching plan corresponding to the student group according to the learning pattern tendency data, English proficiency level data and interest point weighted graph models of the multiple students included in the student group; For each student group, teaching resources are retrieved from the cloud-based teaching resource library according to the basic teaching plan corresponding to the student group, and the corresponding teaching resources are pushed to the student terminal corresponding to the student group according to the teaching progress of the basic teaching plan of the student group; the information acquisition module is used to obtain real-time relevant information, wherein the real-time relevant information includes at least campus activity information and / or real-time news; the teaching management module is also used to generate supplementary push content corresponding to each student group based on real-time relevant information and the learning pattern tendency data, English proficiency level data and interest point weight graph model of multiple students included in the student group using semantic understanding technology, and push the supplementary push content to the student terminal corresponding to the student group.

[0015] Compared with the existing technology, the English teaching management method and system based on semantic understanding and cloud services provided by the present invention have at least the following beneficial effects:

[0016] 1. By collecting data on students' learning habits, English proficiency levels, and points of interest, we can understand the needs of each student more accurately. Based on this data, students are grouped and a basic teaching plan is developed for each group, making the teaching more tailored to the students' actual situation and improving the personalization and pertinence of the teaching. The establishment of a cloud-based teaching resource library enables the centralized management and efficient use of teaching resources. According to the basic teaching plan of each student group, appropriate teaching resources can be automatically retrieved from the resource library, avoiding waste of resources and duplication of work. By pushing teaching resources that meet the basic teaching plan to different student groups, it can ensure that each student can obtain teaching content that matches their ability level, thereby effectively improving learning outcomes. The introduction of real-time relevant information, such as campus activity information and real-time news, can enrich teaching content and enhance students' interest and participation in learning.

[0017] 2. By integrating test data from multiple aspects, including vocabulary, reading, speaking, and writing, a more comprehensive assessment of students' comprehensive English proficiency can be achieved, avoiding the one-sidedness brought about by a single test. Variational modal decomposition of students' oral pronunciation test audio can accurately extract students' key sound features and pronunciation key features, helping students and teachers quickly identify pronunciation problems, such as deficiencies in pitch, volume, and speaking speed. By comparing students' pronunciation key features with those of standard pronunciation, students' oral pronunciation level can be objectively and quantitatively assessed, providing a scientific basis for teaching. Using vocabulary recognition, grammar recognition, and emotion recognition models, students' oral vocabulary features, grammatical features, and conversational emotional features can be further analyzed to comprehensively assess their oral conversational level. This helps students improve the fluency, accuracy, and authenticity of their oral expression.

[0018] 3. Based on the students' interest point test data, establish the students' initial interest point weight map model; determine similar test student groups based on the students' interest point test data and the interest point test data of multiple test students included in each test student group; based on the interest point weight map model of the sample students corresponding to the similar test student groups, complete the students' initial interest point weight map model, and establish a more accurate and complete student interest point weight map model, providing more accurate data support for subsequent grouping and the generation of English teaching plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] This specification will be further described in the form of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting, and in these embodiments, like numbers represent like structures, wherein:

[0020] Figure 1 is a flowchart of an English teaching management method based on semantic understanding and cloud services according to some embodiments of this specification;

[0021] Figure 2 is a schematic diagram of a point of interest weight map model according to some embodiments of this specification;

[0022] Figure 3 This is a module diagram of an English teaching management system based on semantic understanding and cloud services according to some embodiments of this specification. DETAILED DESCRIPTION

[0023] To more clearly illustrate the technical solutions of the embodiments of this specification, the following briefly describes the drawings required for describing the embodiments. Obviously, the drawings described below are merely examples or embodiments of this specification. Those skilled in the art can apply this specification to other similar scenarios based on these drawings without inventive effort. Unless otherwise apparent from the context or otherwise noted, the same reference numerals in the figures represent the same structure or operation.

[0024] Figure 1 is a flowchart of an English teaching management method based on semantic understanding and cloud services according to some embodiments of this specification, such as Figure 1 As shown, an English teaching management method based on semantic understanding and cloud services may include the following steps.

[0025] Step 110: Establish a cloud-based teaching resource library.

[0026] Specifically, the Cloud Teaching Resource Library is a collection of various resources and tools designed to help teachers teach English more effectively while also providing students with richer and more diverse learning materials. The Cloud Teaching Resource Library can be run on a cloud service.

[0027] As an example only, the cloud-based teaching resource library may include the following resources:

[0028] 1. Textbooks and course materials:

[0029] Official English textbooks: including various English textbooks published at home and abroad, such as Oxford English, New Concept English, etc.

[0030] Self-compiled or customized teaching materials: teaching materials written according to specific teaching objectives and student needs.

[0031] Syllabus and lesson plan: A document that details teaching objectives, teaching steps, assessment methods, etc.

[0032] 2. Multimedia teaching resources:

[0033] Video materials: such as English movies, documentaries, TED talks, English teaching videos, etc.

[0034] Audio materials: English songs, podcasts, listening practice materials, phonetics teaching materials, etc.

[0035] Interactive software and applications: online English learning platforms, grammar practice software, pronunciation training tools, etc.

[0036] 3. Reading and Writing Resources:

[0037] Extracurricular reading materials: English novels, essays, news reports, popular science articles, etc.

[0038] Writing Templates and Examples: Templates for various English writing genres, as well as excellent examples of English writing.

[0039] Vocabulary and grammar workbook: Contains materials for vocabulary memorization, grammar explanation and practice.

[0040] 4. Listening and speaking resources:

[0041] Listening training materials: conversation recordings, news broadcasts, telephone conversation simulations, etc.

[0042] Oral practice activities: interactive oral practice such as role-playing, group discussion, and debate.

[0043] Pronunciation and Intonation Training: Professional guidance and practice on English pronunciation and intonation.

[0044] 5. Testing and evaluation tools:

[0045] Unit tests and final exams: exams used to test students’ learning outcomes at a specific stage.

[0046] Proficiency test: sample questions and analysis of international standard English tests such as TOEFL and IELTS.

[0047] Speaking and Listening Assessment Software: Online testing tools that automatically score and provide feedback.

[0048] Step 120: Obtain learning pattern tendency data, English proficiency level data, and interest point weight graph models of multiple students.

[0049] In some embodiments, step 120 specifically includes:

[0050] Obtain English learning test data of multiple students under various teaching methods, and generate students' learning pattern tendency data based on the students' English learning test data;

[0051] Acquire vocabulary test data, reading test data, oral test data, and writing test data of multiple students, and generate students' English proficiency level data based on the students' vocabulary test data, reading test data, oral test data, and writing test data, wherein the students' oral test data includes oral pronunciation test audio and oral dialogue test audio, and can display an English passage to be read aloud, and use audio acquisition equipment to listen to the oral pronunciation test audio of the English passage read aloud by the student, and can conduct oral dialogues with the students through an oral dialogue model, and obtain oral dialogue test audio during the oral dialogue process through the audio acquisition equipment;

[0052] Establish a weighted graph model of interest points for multiple sample students;

[0053] Obtain interest point test data of multiple students, and establish a student interest point weight map based on the student interest point test data and the interest point weight map model of multiple sample students. For example, the student interest point test data can be obtained through a questionnaire.

[0054] As an example only, the following steps may be used to obtain English learning test data of multiple students under various teaching methods, and to generate students' learning pattern tendency data based on the students' English learning test data:

[0055] S11. Determine the test content and method:

[0056] English test questions are designed to cover listening, speaking, reading, writing and other aspects to ensure that students' English proficiency can be comprehensively assessed.

[0057] Design corresponding test scenarios and tasks based on different teaching methods (such as traditional lectures, group discussions, project-based learning, etc.).

[0058] S12. Organization test:

[0059] English learning tests were conducted on multiple students under different teaching methods.

[0060] Ensure that the testing process is standardized to reduce errors.

[0061] S13. Collect data:

[0062] Record students' performance in various tests, including scores, answering time, error types, etc.

[0063] S14. Data preprocessing:

[0064] Clean and organize the collected data to remove invalid or abnormal data.

[0065] The data were normalized for subsequent analysis.

[0066] S15. Learning pattern tendency analysis:

[0067] Analyze students' learning pattern tendencies based on their test performance under different teaching methods.

[0068] For example, by observing students' participation and performance in group discussions, we can determine whether they are inclined to cooperative learning; by analyzing students' completion in project-based learning, we can evaluate their ability to learn independently and solve problems.

[0069] S16. Generate learning mode tendency data:

[0070] Use data analysis tools (such as SPSS, Excel, etc.) to quantify students' learning pattern tendencies.

[0071] Based on the quantitative results, students are categorized into different learning style preferences, such as autonomous learners, cooperative learners, and inquiry learners. Autonomous learners tend to learn independently and enjoy exploring new knowledge on their own. In tests, they may demonstrate strong problem-solving skills and a high accuracy rate. Cooperative learners excel at collaborating with others, deepening their understanding through discussion and exchange. In group discussions or team projects, they may demonstrate high levels of participation and collaboration. Inquiry learners enjoy exploring new knowledge through practice, focusing on hands-on activities and experimental verification. In project-based learning, they may demonstrate strong innovation and problem-solving abilities.

[0072] The vocabulary test data, reading test data, oral test data, and writing test data of multiple students can be obtained in the following ways, and the students' English proficiency level data can be generated based on the students' vocabulary test data, reading test data, oral test data, and writing test data:

[0073] 1. Vocabulary test data

[0074] Testing tools: Use standardized vocabulary testing tools such as vocabulary checklists or online testing platforms.

[0075] How to test: Have students identify or explain the meaning of a given word, or ask them to choose a word from a group of words that matches a specific definition or picture.

[0076] Data Recording: Record the number of words students correctly identified and their performance at different difficulty levels.

[0077] 2. Read the test data

[0078] Test Materials: Select reading materials that are appropriate to the students' level and cover different types of texts (such as stories, expository texts, argumentative essays, etc.).

[0079] Testing method: Set reading comprehension questions and require students to answer questions about the content of the article, or perform information extraction and inference judgment.

[0080] Data recording: Record students’ answering speed and accuracy, as well as their performance on different types of questions.

[0081] 3. Oral test data

[0082] Test format: role-playing, impromptu speech, dialogue simulation, etc.

[0083] Scoring criteria: Evaluation will be based on pronunciation, fluency, grammatical accuracy, vocabulary richness, language organization ability and other aspects.

[0084] Data recording: Use recording equipment to record students' oral performance, which is then scored by the teacher or the oral scoring model according to the scoring criteria.

[0085] 4. Writing test data

[0086] Test task: Assign writing topics and require students to write short essays, argumentative essays or stories, etc.

[0087] Scoring criteria: Evaluation will be based on content quality, clarity of structure, accuracy of language expression, vocabulary richness and creativity.

[0088] Data Recording: Student written work is collected and scored by the teacher or writing scoring model according to the rubric.

[0089] Integrate student performance data on vocabulary, reading, speaking, and writing tests into a single database. If the data from different tests differ in scale (e.g., vocabulary tests are presented as numbers, while speaking tests are presented as scores), standardize the data to facilitate comparison and comprehensive analysis. Assign appropriate weights to each test based on its importance and the need for a comprehensive assessment of the student's English proficiency. Based on this weighting and the student's performance on each test, calculate the student's overall English proficiency score.

[0090] Preferably, the student's English proficiency data is generated based on the student's vocabulary test data, reading test data, oral test data, and writing test data, including:

[0091] Perform variational modal decomposition on students' oral pronunciation test audio to extract students' key sound features and pronunciation key features;

[0092] Based on the students' key vocal features, similar standard spoken pronunciation audio is obtained from the standard spoken pronunciation audio library. The standard spoken pronunciation audio library includes standard spoken pronunciation audio produced by different professional language teachers and phoneticians, and contains clear and accurate oral pronunciation demonstrations of English passages to be read aloud;

[0093] Perform variational mode decomposition on audio similar to standard spoken pronunciation to extract key features of standard pronunciation;

[0094] Determine the student's oral pronunciation level based on the student's pronunciation key features and the standard pronunciation key features;

[0095] Using a vocabulary recognition model to test audio of students’ spoken conversations, the student’s spoken vocabulary characteristics are determined;

[0096] Using the grammar recognition model to test the students’ spoken conversations, the grammar characteristics of the students’ spoken language are determined.

[0097] Using emotion recognition models to test audio of students’ spoken conversations, we can determine the emotional characteristics of students’ conversations.

[0098] The student's oral conversation level is determined based on the student's oral vocabulary characteristics, oral grammar characteristics and conversation emotion characteristics, wherein the student's English proficiency level data at least includes the student's oral pronunciation level and oral conversation level.

[0099] Specifically, the key voice features of students can be extracted according to the following formula:

[0100] S21. For each standard spoken pronunciation audio, perform variational mode decomposition on the standard spoken pronunciation audio to generate multiple eigenmode functions corresponding to the standard spoken pronunciation audio;

[0101] S22. Determine multiple candidate variational mode decomposition factors (e.g., center frequency mean, center frequency standard deviation, bandwidth mean, bandwidth standard deviation, instantaneous frequency change rate mean, instantaneous frequency change rate standard deviation, etc.);

[0102] S23. For each standard spoken pronunciation audio, determine the eigenvalue corresponding to each candidate variational mode decomposition factor according to multiple eigenmode functions corresponding to the standard spoken pronunciation audio;

[0103] S24. For each candidate variational mode decomposition factor, calculate the factor difference value corresponding to the candidate variational mode decomposition factor according to the eigenvalue of each standard spoken pronunciation audio corresponding to the candidate variational mode decomposition factor, and select the candidate variational mode decomposition factor whose factor difference value is greater than the factor difference value threshold as the key variational mode decomposition factor;

[0104] S25. performing variational mode decomposition on the student's oral pronunciation test audio to generate a plurality of eigenmode functions corresponding to the student's oral pronunciation test audio;

[0105] S26. Determine the eigenvalue corresponding to each key variational mode decomposition factor based on multiple eigenmode functions corresponding to the student's oral pronunciation test audio, wherein the student's key sound features may include the eigenvalue corresponding to each key variational mode decomposition factor.

[0106] For example, the factor difference value corresponding to the candidate variational mode decomposition factor can be calculated according to the following formula:

[0107] ,

[0108] in, For the The factor difference values ​​corresponding to the candidate variational mode decomposition factors, is the eigenvalue of the i-th candidate variational mode decomposition factor corresponding to the n-th standard spoken pronunciation audio, The total number of standard spoken pronunciation audios.

[0109] The sound similarity between the standard spoken pronunciation audio and the student's spoken pronunciation test audio can be calculated according to the following formula:

[0110] ,

[0111] in, is the sound similarity between the mth standard spoken pronunciation audio and the student's spoken pronunciation test audio, is the eigenvalue of the i-th key variational mode decomposition factor corresponding to the m-th standard spoken pronunciation audio, The eigenvalue of the i-th key variational mode decomposition factor corresponding to the student's oral pronunciation test audio, is the total number of key variational mode decomposition factors.

[0112] The standard spoken pronunciation audio with the greatest sound similarity may be used as the similar standard spoken pronunciation audio.

[0113] Specifically, multiple intrinsic mode function characteristic factors (e.g., center frequency, bandwidth, energy distribution, amplitude, phase, instantaneous frequency, etc.) can be determined. Based on these multiple intrinsic mode function characteristic factors, key pronunciation features are extracted from the student's oral pronunciation test audio. These key pronunciation features can include the eigenvalues ​​of each intrinsic mode function characteristic factor corresponding to each intrinsic mode function of the student's oral pronunciation test audio. Based on these multiple intrinsic mode function characteristic factors, key features of standard pronunciation are extracted from similar standard spoken pronunciation audio. These key features of standard pronunciation can include the eigenvalues ​​of each intrinsic mode function characteristic factor corresponding to each intrinsic mode function of the similar standard spoken pronunciation audio.

[0114] The student's spoken pronunciation level can be determined based on the student's pronunciation key features and standard pronunciation key features using a spoken pronunciation level recognition model. The spoken pronunciation level recognition model can be a convolutional neural network model.

[0115] Specifically, variational mode decomposition (VMD) can decompose spoken pronunciation test audio into multiple intrinsic mode functions (IMFs), each of which corresponds to a specific frequency component or sound feature. By determining key VMD factors and identifying similar standard spoken pronunciation audio based on these factors, it is possible to identify standard spoken pronunciation audio that closely resembles the student's timbre. This reduces interference information in subsequent spoken pronunciation assessments and improves the accuracy of spoken pronunciation assessments.

[0116] Furthermore, through variational modal decomposition, key vocal and pronunciation features of students can be precisely extracted, such as the clarity of vowels, the force of consonants, and the ebb and flow of intonation. The standard spoken pronunciation audio library consists of clear and accurate audio recordings of standard spoken language produced by various professional language teachers and phoneticians, serving as a benchmark for evaluation. Comparative analysis with standard pronunciation audio allows for an objective and accurate assessment of students' spoken pronunciation, making it easier to identify pronunciation issues, such as inaccurate pronunciation of certain vowels or overly or under-pronounced consonants.

[0117] The oral vocabulary recognition model lies between acoustic signal segmentation and conceptual access. It connects the segmented acoustic signal with vocabulary in long-term memory, leading to vocabulary recognition. In the students' oral conversation test audio, the vocabulary recognition model identifies, categorizes, and counts the vocabulary used. This model can calculate the total vocabulary used by the students in the oral conversation, as well as the number of different vocabulary categories (such as nouns, verbs, and adjectives). This model analyzes the richness and diversity of the students' vocabulary, including the inclusion of advanced vocabulary or specialized terminology. It also assesses the accuracy of the vocabulary used, and any misuse or confusion.

[0118] The grammar recognition model analyzes the sentence structure, grammatical rules, and word order in students' spoken conversations. Using tools like syntax trees and dependency trees, it converts students' spoken conversations into grammatically clear sentences for further analysis. It analyzes the sentence types (e.g., simple vs. complex) and the complexity of the sentence structures used. It assesses whether students adhere to correct grammatical rules, such as tense, voice, and subject-verb agreement. It also checks whether students' word order is correct and conforms to Chinese grammatical conventions.

[0119] Emotion recognition models can analyze emotional expressions in students' spoken conversations, such as nervousness, calmness, excitement, and embarrassment. Using vocabulary-based methods (such as sentiment lexicons), sentence structure analysis, dependency analysis, and machine learning algorithms, these models can identify students' emotional tendencies in conversations. The spoken vocabulary recognition model, grammar recognition model, and emotion recognition model can all be long-short-term memory network models.

[0120] The spoken language evaluation model can be used to determine the student's spoken conversation level based on the spoken vocabulary features output by the spoken vocabulary recognition model, the spoken grammatical features output by the grammar recognition model, and the dialogue emotion features output by the emotion recognition model. The spoken language evaluation model can be a convolutional neural network model.

[0121] In some embodiments, establishing a weighted graph model of interest points for multiple sample students includes:

[0122] Obtaining test data of interest points of multiple test students;

[0123] Dividing the multiple test students into multiple test student groups according to the test data of the multiple test students' points of interest;

[0124] For each test student group, based on the interest point test data of multiple test students included in the test student group, an interest point weight graph model of the sample students corresponding to the test student group is established.

[0125] Specifically, first, it's necessary to clarify the specific meaning and scope of "interests." These can include subject areas (such as mathematics, science, literature, and art), extracurricular activities (such as sports, music, and programming), specific topics (such as environmental protection and space exploration), or any other areas related to student interests. Based on the definition of interests, design a questionnaire or scale containing a series of questions or statements related to these interests. Questions can be multiple-choice, ranking, or rating questions to test students' preferences for different interests. Distribute the designed questionnaire to students through channels such as schools, classes, and online platforms. Ensure that the questionnaire reaches all student groups to be tested. The collected questionnaire data should be entered into a spreadsheet or database for subsequent analysis and processing. During data entry, ensure data accuracy and completeness. Check the data for missing values, outliers, and duplicates. For missing values, interpolation, mean substitution, or deletion can be used to address them. For outliers, determine whether they truly reflect student interests or are the result of operational errors or data entry errors. The preference degree of the test students for different points of interest is quantified into numerical values ​​to determine the preference values ​​of the test students for different points of interest.

[0126] For any two test students, the similarity of the interest preferences of the two test students can be calculated based on their preference values ​​for different interest points.

[0127] For example, the similarity of interest preferences of two test students can be calculated according to the following formula:

[0128] ,

[0129] in, is the similarity of interest preferences between the i-th test student and the j-th test student, is the similarity of the interest preference of the i-th test student to the k-th interest point, is the similarity of the interest preference of the jth test student to the kth interest point, is the total number of points of interest.

[0130] pass The mean clustering algorithm clusters multiple test students according to the similarity of interest preferences of any two test students, and divides multiple test students into multiple test student groups.

[0131] For each test student group, the weight corresponding to each interest point can be calculated according to the following formula:

[0132] ,

[0133] in, is the weight corresponding to the g-th interest point, is the preference value of the e-th test student in the test student group for the g-th interest point, is the total number of test students included in the test student group, is the total number of points of interest.

[0134] For example only, Figure 2 is a schematic diagram of a point of interest weight map model according to some embodiments of this specification, such as Figure 2 As shown, the interest point weight graph model may include nodes representing interest points and nodes representing students. The nodes representing interest points and the nodes representing students are connected by edges, and the weight of the edge represents the weight corresponding to the interest point.

[0135] In some embodiments, establishing a student's interest point weighted graph model based on the student's interest point test data and the interest point weighted graph models of multiple sample students includes:

[0136] According to the students’ interest point test data, establish the students’ initial interest point weight graph model;

[0137] Determine similar test student groups based on the student's interest point test data and the interest point test data of multiple test students included in each test student group;

[0138] Based on the interest point weighted graph model of the sample students corresponding to the similar test student group, the student's initial interest point weighted graph model is completed to establish the student's interest point weighted graph model.

[0139] Specifically, based on the student's interest point test data, the student's preference values ​​for different interest points are determined, and interest points with preference values ​​greater than a preference value threshold are selected as the student's target interest points. Based on the student's preference value for the target interest point and the average preference values ​​of the multiple test students included in the test student group for the target interest point, the matching value of the test student group is calculated, and the test student group with a matching value greater than the matching value threshold is selected as a similar test student group. The matching value calculation method is similar to the method for calculating the similarity of the interest preferences of two test students and will not be repeated here.

[0140] The weighted graph completion model is used to complete the student's initial interest point weighted graph model based on the interest point weighted graph model of the sample students corresponding to the similar test student group and the matching value of each test student group, and the student's interest point weighted graph model is established, wherein the weighted graph completion model can be a convolutional neural network model.

[0141] Step 130 : grouping the students according to their learning pattern tendency data, English proficiency level data, and interest point weight graph model to determine a plurality of student groups.

[0142] In some embodiments, step 130 specifically includes:

[0143] For any two students, the similarity of the weighted graph models of the two students' points of interest is calculated based on their weighted graph models of the two students' points of interest;

[0144] According to the similarity of the weighted graph model of the points of interest of any two students, the multiple students are first grouped to determine multiple first student groups. For example, a K-means clustering algorithm can be used to perform the first grouping of the multiple students according to the similarity of the weighted graph model of the points of interest of any two students to determine multiple first student groups.

[0145] For each first student group, based on the English proficiency level data of any two students included in the first student group, calculate the similarity of the English proficiency levels of any two students, and based on the similarity of the English proficiency levels of any two students, perform a second grouping on the multiple students included in the first student group to determine multiple second student groups. For example, a K-means clustering algorithm can be used to perform a second grouping on the multiple students included in the first student group based on the similarity of the English proficiency levels of any two students included in the first student group to determine multiple second student groups.

[0146] For each second student group, based on the learning pattern tendency data of any two students included in the second student group, the similarity of the learning pattern tendencies of any two students is calculated. Based on the similarity of the learning pattern tendencies of any two students, the multiple students included in the second student group are grouped for a third time to determine multiple student groups. For example, the K-means clustering algorithm can be used to group the multiple students included in the second student group for a third time to determine multiple student groups based on the similarity of the learning pattern tendencies of any two students included in the second student group.

[0147] For example, the similarity of the interest point weight graph models of two students can be calculated according to the following formula:

[0148] ,

[0149] in, is the similarity between the weighted graph model of interest points of the i-th student and the j-th student, is the weight corresponding to the gth interest point in the interest point weight graph model of the i-th student, is the weight corresponding to the gth interest point in the interest point weight graph model of the jth student.

[0150] Step 140 : For each student group, a basic teaching plan corresponding to the student group is generated based on the learning pattern tendency data, English proficiency level data, and interest point weight graph model of the multiple students included in the student group.

[0151] In some embodiments, step 140 specifically includes:

[0152] Determine the target teaching method corresponding to the student group based on the learning pattern tendency data of multiple students included in the student group;

[0153] Determine multiple target English knowledge points corresponding to the student group based on English proficiency level data of multiple students included in the student group;

[0154] Based on the interest point weighted graph model of multiple students included in the student group and multiple target English knowledge points, a basic teaching plan corresponding to the student group is generated, wherein the basic teaching plan at least includes target English knowledge points, target teaching methods and teaching objectives at multiple teaching time nodes.

[0155] Specifically, the most appropriate teaching method for the entire student group is determined based on the learning style preferences of each member of the group (i.e., multiple student clients). Learning style preferences may include visual learners, auditory learners, hands-on learners, and other types. For example, if the majority of students in a group are visual learners, the teaching method may focus on visual materials such as charts and videos.

[0156] Based on the English proficiency data of each student in the student group, the specific English knowledge points that they need to master or improve together are determined. These knowledge points may include grammar rules, vocabulary expansion, reading comprehension skills, etc. By analyzing the students' ability levels, the system can identify the common weaknesses of the student group or areas that need to be strengthened, and then use these areas as target English knowledge points. After determining the target teaching method and target English knowledge points,

[0157] A basic teaching plan is generated using a plan generation model based on the weighted graph model of students' points of interest and these knowledge points. By integrating students' points of interest with target knowledge points, a teaching plan is designed that both meets learning objectives and stimulates student interest. This plan includes at least the target English knowledge points scheduled at different teaching time points, the corresponding teaching methods, and the desired teaching objectives. The plan generation model can be a convolutional neural network model.

[0158] Step 150: For each student group, teaching resources are retrieved from the cloud teaching resource library according to the basic teaching plan corresponding to the student group, and the corresponding teaching resources are pushed to the student terminal corresponding to the student group according to the teaching progress of the basic teaching plan of the student group.

[0159] Specifically, a reasonable push plan is developed based on the teaching progress of the student group's basic teaching plan. The push plan will clearly define the teaching resources to be pushed at each teaching time point, as well as their type and quantity. According to the push plan, the corresponding teaching resources are automatically pushed to the student end, ensuring that students have timely access to the required learning materials.

[0160] Step 160: Obtain real-time relevant information.

[0161] The real-time related information includes at least campus activity information and / or real-time news.

[0162] In some embodiments, step 160 specifically includes:

[0163] Determine a first target website, and obtain candidate campus activity information from the first target website (e.g., a campus official website) using crawler technology;

[0164] Determine a second target website, and obtain candidate real-time news from the second target website (e.g., a news publishing website) based on crawler technology;

[0165] Screen candidate campus activity information and candidate real-time news to determine valid campus activity information and valid real-time news.

[0166] In some embodiments, screening candidate campus activity information and candidate real-time news to determine valid campus activity information and valid real-time news includes:

[0167] For each student group, generating a group interest point weighted map model corresponding to the student group based on the interest point weighted map models of the multiple students included in the student group, and performing weighted summation on the interest point weighted map models of the multiple students included in the student group to generate a group interest point weighted map model corresponding to the student group;

[0168] Determine points of interest corresponding to candidate campus activity information;

[0169] Determine points of interest corresponding to candidate real-time news;

[0170] According to the group interest point weight graph model corresponding to each student group, the interest points corresponding to the candidate campus activity information and the interest points corresponding to the candidate real-time news, the candidate campus activity information and the candidate real-time news are screened to determine the valid campus activity information and the valid real-time news.

[0171] Specifically, determine whether there is at least one student group whose corresponding interest points of the candidate campus activity information and the corresponding interest points of the candidate real-time news in the group interest point weight graph model are greater than the weight threshold. If so, determine them as valid campus activity information and valid real-time news.

[0172] Step 170: For each student group, based on real-time relevant information and the learning pattern tendency data, English proficiency level data and interest point weight graph model of multiple students included in the student group, semantic understanding technology is used to generate supplementary push content corresponding to the student group, and the supplementary push content is pushed to the student end corresponding to the student group.

[0173] In some embodiments, step 170 specifically includes:

[0174] For each student group, based on the points of interest corresponding to the valid campus activity information or the points of interest corresponding to the valid real-time news, the target campus activity information or target real-time news is determined, and based on the English proficiency level data of the multiple students included in the student group using semantic understanding technology, English reading articles are generated according to the target campus activity information or target real-time news. Based on the learning pattern tendency data of the multiple students included in the student group, the English reading articles are converted into supplementary push content, and the supplementary push content is pushed to the student end corresponding to the student group.

[0175] Specifically, advanced semantic understanding technology is used to process targeted campus event information or real-time news. This technology deeply analyzes text content, extracts key information, and generates coherent, logically structured English reading passages based on this information. When generating passages, the student group's English proficiency data is fully considered. The difficulty of the passages is adjusted based on factors such as students' vocabulary and grammar proficiency, ensuring that the passages are both challenging and easy to understand.

[0176] After generating English reading passages, we transform them into supplementary content tailored to different learning styles based on the learning preferences of the student groups. For example, for visual learners, we transform the passages into charts, images, or videos; for auditory learners, we might provide audio readings.

[0177] The generated supplementary push content will be pushed to the corresponding student terminals of the student group through appropriate means (e.g., online download, email, in-app notification, etc.). The push process should ensure the timeliness and accuracy of the content so that students can access and read these supplementary materials in a timely manner.

[0178] Figure 3 This is a module diagram of an English teaching management system based on semantic understanding and cloud services according to some embodiments of this specification. Figure 3As shown, an English teaching management system based on semantic understanding and cloud services may include a resource management module, a teaching management module and an information acquisition module.

[0179] Resource management module, used to establish a cloud-based teaching resource library;

[0180] The teaching management module is used to obtain the learning pattern tendency data, English proficiency level data and interest point weighted graph model of multiple students; group the multiple students according to the learning pattern tendency data, English proficiency level data and interest point weighted graph model of the multiple students to determine multiple student groups; for each student group, generate a basic teaching plan corresponding to the student group according to the learning pattern tendency data, English proficiency level data and interest point weighted graph model of the multiple students included in the student group; for each student group, retrieve teaching resources from the cloud teaching resource library according to the basic teaching plan corresponding to the student group, and push the corresponding teaching resources to the student terminal corresponding to the student group according to the teaching progress of the basic teaching plan of the student group;

[0181] An information acquisition module, configured to acquire real-time relevant information, wherein the real-time relevant information includes at least campus activity information and / or real-time news;

[0182] The teaching management module is also used to generate supplementary push content corresponding to each student group based on real-time relevant information and the learning pattern tendency data, English proficiency level data and interest point weight graph model of multiple students included in the student group using semantic understanding technology, and push the supplementary push content to the student end corresponding to the student group.

[0183] An English teaching management system based on semantic understanding and cloud services can be used to implement an English teaching management method based on semantic understanding and cloud services, which will not be described in detail here.

[0184] Finally, it should be understood that the embodiments described in this specification are intended only to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, by way of example and not limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly described and illustrated in this specification.

Claims

1. An English teaching management method based on semantic understanding and cloud services, characterized in that: include: Establish a cloud-based teaching resource library; Obtaining learning pattern tendency data, English proficiency level data, and a weighted graph model of points of interest for multiple students, wherein the weighted graph model of points of interest includes nodes representing points of interest and nodes representing students, the nodes representing points of interest and the nodes representing students are connected by edges, and the weights of the edges represent the weights corresponding to the points of interest; Grouping the plurality of students according to the learning pattern tendency data, English proficiency level data, and interest point weight graph model to determine a plurality of student groups; For each student group, a basic teaching plan corresponding to the student group is generated based on the learning pattern tendency data, English proficiency level data, and interest point weight graph model of the multiple students included in the student group; For each student group, teaching resources are retrieved from the cloud teaching resource library according to the basic teaching plan corresponding to the student group, and the corresponding teaching resources are pushed to the student terminal corresponding to the student group according to the teaching progress of the basic teaching plan of the student group; Acquiring real-time related information, wherein the real-time related information includes at least campus activity information and / or real-time news; For each student group, based on real-time relevant information and the learning pattern tendency data, English proficiency data, and interest point weighted graph model of the multiple students included in the student group, semantic understanding technology is used to generate supplementary push content corresponding to the student group, and the supplementary push content is pushed to the student terminals corresponding to the student group; Among them, the English proficiency data of multiple students are obtained, including: Obtain vocabulary test data, reading test data, oral test data, and writing test data for multiple students, and generate students' English proficiency data based on the students' vocabulary test data, reading test data, oral test data, and writing test data, specifically including: Perform variational modal decomposition on students' oral pronunciation test audio to extract students' key sound features and pronunciation key features; Based on the key sound features of students, similar standard spoken pronunciation audio is obtained from the standard spoken pronunciation audio library; Perform variational mode decomposition on audio similar to standard spoken pronunciation to extract key features of standard pronunciation; Determine the student's oral pronunciation level based on the student's pronunciation key features and the standard pronunciation key features; Extract key voice features of students, including: S21. For each standard spoken pronunciation audio, perform variational mode decomposition on the standard spoken pronunciation audio to generate multiple eigenmode functions corresponding to the standard spoken pronunciation audio; S22. Determine multiple candidate variational mode decomposition factors, including center frequency mean, center frequency standard deviation, bandwidth mean, bandwidth standard deviation, instantaneous frequency change rate mean, and instantaneous frequency change rate standard deviation; S23. For each standard spoken pronunciation audio, determine the eigenvalue corresponding to each candidate variational mode decomposition factor according to multiple eigenmode functions corresponding to the standard spoken pronunciation audio; S24. For each candidate variational mode decomposition factor, calculate the factor difference value corresponding to the candidate variational mode decomposition factor according to the eigenvalue of the candidate variational mode decomposition factor corresponding to each standard spoken pronunciation audio, and take the candidate variational mode decomposition factor with a factor difference value greater than the factor difference value threshold as the key variational mode decomposition factor. Calculate the factor difference value corresponding to the candidate variational mode decomposition factor according to the following formula: ， in, is the factor difference value corresponding to the i-th candidate variational mode decomposition factor, is the eigenvalue of the i-th candidate variational mode decomposition factor corresponding to the n-th standard spoken pronunciation audio, The total number of standard spoken pronunciation audios; S25. performing variational mode decomposition on the student's oral pronunciation test audio to generate a plurality of eigenmode functions corresponding to the student's oral pronunciation test audio; S26. Determine the eigenvalue corresponding to each key variational mode decomposition factor based on multiple eigenmode functions corresponding to the student's oral pronunciation test audio, wherein the student's key sound features include the eigenvalue corresponding to each key variational mode decomposition factor.

2. The English teaching management method based on semantic understanding and cloud services according to claim 1, characterized in that: Obtain learning pattern tendency data, English proficiency data, and interest point weight graph models for multiple students, including: Obtain English learning test data of multiple students under various teaching methods, and generate students' learning pattern tendency data based on the students' English learning test data; Establish a weighted graph model of interest points for multiple sample students; Obtain interest point test data of multiple students, and establish a student interest point weighted graph model based on the student interest point test data and the interest point weighted graph models of multiple sample students.

3. The English teaching management method based on semantic understanding and cloud services according to claim 2 is characterized in that: Establish a weighted graph model of interest points for multiple sample students, including: Obtain interest point test data of multiple test students; Dividing the multiple test students into multiple test student groups according to the test data of the multiple test students' points of interest; For each test student group, based on the interest point test data of multiple test students included in the test student group, a weighted graph model of interest points of sample students corresponding to the test student group is established; Based on the students' interest point test data and the interest point weighted graph models of multiple sample students, a student interest point weighted graph model is established, including: According to the students’ interest point test data, establish the students’ initial interest point weight graph model; Determine similar test student groups based on the student's interest point test data and the interest point test data of multiple test students included in each test student group; Based on the interest point weighted graph model of the sample students corresponding to the similar test student group, the student's initial interest point weighted graph model is completed to establish the student's interest point weighted graph model.

4. The English teaching management method based on semantic understanding and cloud services according to claim 2, characterized in that: The student's oral test data includes oral pronunciation test audio and oral dialogue test audio; Generate students' English proficiency data based on their vocabulary test data, reading test data, speaking test data, and writing test data, including: using a vocabulary recognition model to determine students' oral vocabulary characteristics based on their oral dialogue test audio; Using the grammar recognition model to test the students’ spoken conversations, the grammar characteristics of the students’ spoken language are determined. Using emotion recognition models to test audio of students’ spoken conversations, we can determine the emotional characteristics of students’ conversations. The student's oral conversation level is determined based on the student's oral vocabulary characteristics, oral grammar characteristics and conversation emotion characteristics, wherein the student's English proficiency level data at least includes the student's oral pronunciation level and oral conversation level.

5. The English teaching management method based on semantic understanding and cloud services according to any one of claims 2 to 4, characterized in that: The plurality of students are grouped according to the learning pattern tendency data, English proficiency level data, and the interest point weight graph model to determine a plurality of student groups, including: For any two students, the similarity of the weighted graph models of the two students' points of interest is calculated based on their weighted graph models of the two students' points of interest; According to the similarity of the weighted graph model of the points of interest of any two students, multiple students are grouped for the first time to determine multiple first student groups; For each first student group, calculating the similarity of the English proficiency levels of any two students included in the first student group based on the English proficiency level data of the two students, and performing a second grouping of the multiple students included in the first student group based on the similarity of the English proficiency levels of the two students to determine multiple second student groups; For each second student group, the learning pattern tendency similarity of any two students included in the second student group is calculated based on the learning pattern tendency data of any two students. Based on the learning pattern tendency similarity of any two students, the multiple students included in the second student group are grouped for the third time to determine multiple student groups.

6. The English teaching management method based on semantic understanding and cloud services according to any one of claims 2 to 4, characterized in that: Generate a basic teaching plan corresponding to the student group based on the learning pattern tendency data, English proficiency level data, and interest point weight graph model of the multiple students included in the student group, including: Determining a target teaching method corresponding to the student group based on the learning pattern tendency data of the multiple student terminals included in the student group; Determining a plurality of target English knowledge points corresponding to the student group based on the English proficiency level data of the plurality of student terminals included in the student group; Based on the interest point weight graph model of multiple students included in the student group and multiple target English knowledge points, a basic teaching plan corresponding to the student group is generated, wherein the basic teaching plan at least includes target English knowledge points, target teaching methods and teaching objectives at multiple teaching time nodes.

7. The English teaching management method based on semantic understanding and cloud services according to any one of claims 1 to 4, characterized in that: Get real-time, relevant information, including: Determine a first target website, and obtain candidate campus activity information from the first target website based on crawler technology; Determine a second target website, and obtain candidate real-time news from the second target website based on crawler technology; Screen candidate campus activity information and candidate real-time news to determine valid campus activity information and valid real-time news.

8. The English teaching management method based on semantic understanding and cloud services according to claim 7, characterized in that: Screen candidate campus event information and candidate real-time news to determine valid campus event information and valid real-time news, including: For each student group, generating a group interest point weighted graph model corresponding to the student group according to the interest point weighted graph models of the multiple students included in the student group; Determine points of interest corresponding to candidate campus activity information; Determine points of interest corresponding to candidate real-time news; According to the group interest point weight graph model corresponding to each student group, the interest points corresponding to the candidate campus activity information and the interest points corresponding to the candidate real-time news, the candidate campus activity information and the candidate real-time news are screened to determine the valid campus activity information and the valid real-time news.

9. The English teaching management method based on semantic understanding and cloud services according to claim 8, characterized in that: Based on real-time relevant information and the learning pattern tendency data, English proficiency data, and interest point weight graph model of multiple students included in the student group, semantic understanding technology is used to generate supplementary push content corresponding to the student group, and the supplementary push content is pushed to the student terminals corresponding to the student group, including: For each student group, based on the points of interest corresponding to valid campus activity information or the points of interest corresponding to valid real-time news, the target campus activity information or target real-time news is determined, and based on the English proficiency level data of multiple students included in the student group using semantic understanding technology, English reading articles are generated according to the target campus activity information or target real-time news. Based on the learning pattern tendency data of multiple students included in the student group, the English reading articles are converted into supplementary push content, and the supplementary push content is pushed to the student terminal corresponding to the student group.

10. An English teaching management system based on semantic understanding and cloud services, applied to an English teaching management method based on semantic understanding and cloud services as claimed in any one of claims 1 to 9, characterized in that: include: Resource management module, used to establish a cloud-based teaching resource library; A teaching management module is used to obtain learning pattern tendency data, English proficiency level data, and interest point weighted graph models of multiple students; group the multiple students according to the learning pattern tendency data, English proficiency level data, and interest point weighted graph models of the multiple students to determine multiple student groups; for each student group, generate a basic teaching plan corresponding to the student group according to the learning pattern tendency data, English proficiency level data, and interest point weighted graph models of the multiple students included in the student group; for each student group, retrieve teaching resources from a cloud-based teaching resource library according to the basic teaching plan corresponding to the student group, and push corresponding teaching resources to the student terminal corresponding to the student group according to the teaching progress of the basic teaching plan of the student group; An information acquisition module, configured to acquire real-time related information, wherein the real-time related information includes at least campus activity information and / or real-time news; The teaching management module is also used to generate supplementary push content corresponding to each student group based on real-time relevant information and the learning pattern tendency data, English proficiency level data and interest point weight graph model of multiple students included in the student group using semantic understanding technology, and push the supplementary push content to the student terminal corresponding to the student group.

Images