Interactive digital teaching system based on AI big data models

Abstract

This application relates to the field of educational informatization technology and discloses an interactive digital teaching system based on an AI big data model, including: a data acquisition module for acquiring standardized real-time interactive data and sending it to an AI big data model processing module; an interactive intent recognition module for extracting feature vectors from student response text and inputting them into the AI ​​big data model processing module; an AI big data model processing module for receiving real-time interactive data and feature vectors, and generating response feedback data and content retrieval instructions based on parsing and intent judgment results; a teaching content storage module for retrieving teaching resources according to content retrieval instructions and pushing them to the interactive module; and an interactive module for receiving student response content, displaying teaching resources and response feedback data, and synchronizing the response content to the data acquisition module. This application improves the accuracy of student intent recognition and the timeliness of dynamic adjustments to teaching content.

Description

Technical Field

[0001] This manual relates to the field of educational information technology, and in particular to interactive digital teaching systems based on AI big data models. Background Technology

[0002] Currently, AI-based large-scale model-based educational systems are being gradually adopted in digital teaching scenarios such as K-12 education and vocational education. These systems leverage the generation and reasoning capabilities of large language models and multimodal processing models to enhance the interactivity and personalization of teaching. However, existing AI teaching systems still suffer from the following technical shortcomings in practical applications: On the one hand, the depth of teaching interaction is insufficient, and the response mode is mechanical. Existing systems mostly adopt a response mode based on pre-stored question banks and keyword matching, without truly combining the deep understanding and generation capabilities of AI large models. They can only match and judge students' standardized answers, making it difficult to identify logical deviations and thinking errors in students' responses, let alone generate customized explanation content for students' individual questions. The interaction is only at a superficial stage of checking the standard answer to the question.

[0003] On the other hand, the teaching content delivery mechanism is rigid and lacks dynamic adaptability. The existing system uses a fixed template or a simple score association mode for delivering teaching content. It cannot dynamically adjust the difficulty, explanation method and delivery pace of teaching content based on real-time data such as students' response speed, error type and question direction during the interaction process. This results in a mismatch between teaching content and students' actual learning needs, making it difficult to achieve personalized teaching.

[0004] Furthermore, the feedback mechanism is simplistic, and the closed-loop teaching process is incomplete. The feedback mechanism of existing interactive teaching systems can only output binary results of correct or incorrect, failing to provide students with personalized error analysis and thought process guidance, nor can it provide teachers with real-time feedback on students' learning gaps. This makes it difficult for interactive teaching to form a complete closed loop of questioning, answering, feedback, and adjustment.

[0005] Therefore, there is an urgent need for an interactive digital teaching system based on large AI models to solve the above problems. Summary of the Invention

[0006] In view of this, the present invention aims to propose an interactive digital teaching system based on a large AI model to solve the problems of low accuracy in recognizing student intentions and poor timeliness in dynamically adjusting teaching content in existing AI teaching systems, thereby improving the accuracy of recognizing student intentions and the timeliness of dynamically adjusting teaching content.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: An interactive digital teaching system based on AI big data models includes: The data acquisition module is used to collect real-time interactive data during student interaction, perform standardized processing, and send it to the AI ​​large model processing module. The interactive intent recognition module is used to extract feature vectors from student response texts and input them into the AI ​​large model processing module; The AI ​​large model processing module is used to receive standardized real-time interactive data and feature vectors, parse the real-time interactive data and feature vectors and determine the intent, and generate response feedback data and content retrieval instructions. The teaching content storage module is used to store teaching resources and, according to the content retrieval instructions output by the AI ​​large model processing module, retrieve the corresponding teaching resources and push them to the interactive module. The interactive module is used to receive the response content input by the student; to display the teaching resources pushed by the teaching content storage module and the response feedback data generated by the AI ​​big model processing module to the student; and to synchronize the response content input by the student to the data acquisition module.

[0008] The beneficial effects of this solution are as follows: Existing AI teaching systems generally suffer from insufficient interactive depth, fixed teaching content delivery, and a single feedback mechanism, which restricts the interactivity and personalization of teaching. This solution combines an interactive intent recognition module with an AI large-scale model processing module to extract feature vectors from student response texts and perform in-depth analysis, thereby improving the accuracy of student intent recognition; it collects student interaction data in real time through a data acquisition module and links it with the teaching content storage module to push teaching resources in real time according to content retrieval commands, improving the timeliness of dynamic adjustments to teaching content; and by constructing a closed-loop linkage between data acquisition, AI processing, and interaction modules, it realizes a complete teaching process of questioning, answering, providing feedback, and adjusting.

[0009] Furthermore, the data acquisition module includes a text acquisition unit, a time acquisition unit, an operation acquisition unit, and a data integration unit; The text acquisition unit is used to acquire text data from the student's response content through the input interface of the interactive module, store it in UTF-8 encoding format, and remove invalid characters from the text data. The time acquisition unit is used to collect the total time from when the student receives the question to when they submit their answer, as well as the pause time during the answering process, in real time through a timer to obtain response time data; The operation acquisition unit is used to collect students' click operations and backtracking operations through the operation monitoring interface of the terminal device, record the operation type and operation time, and obtain behavioral operation data. The data integration unit is used to integrate the data collected by the text acquisition unit, the time acquisition unit and the operation acquisition unit into JSON format data, and send it to the AI ​​large model processing module via HTTP protocol.

[0010] Beneficial effects: By setting up text acquisition units, time acquisition units, and operation acquisition units to collect text data, response time data, and behavioral operation data respectively, and integrating them into JSON format and sending them via HTTP protocol, the structured acquisition and standardized transmission of multi-source interactive data is realized.

[0011] Furthermore, the interactive intent recognition module is built based on the BERT model and includes a feature extraction unit and a feature filtering unit; The feature extraction unit is used to receive the text portion of the standardized real-time interactive data sent by the data acquisition module, extract the semantic features, logical features, and error features from the student's response content, and output a feature vector. The feature filtering unit is used to remove invalid features that are irrelevant to the teaching content, retain core features that are related to knowledge points and questions, and send the filtered feature vectors to the AI ​​large model processing module.

[0012] Beneficial effects: By constructing feature extraction and feature filtering units based on the BERT model, semantic features, logical features, and error features in student responses are extracted, and irrelevant features are filtered out, thereby improving the feature extraction accuracy and parsing efficiency of the interactive intent recognition module.

[0013] Furthermore, the AI ​​large model processing module adopts a large language model based on the Transformer architecture, including an intent parsing unit and a content generation unit; The intent parsing unit is used to receive the feature vector sent by the interactive intent recognition module, and combine it with the response time data and behavioral operation data of the data acquisition module to generate a student interaction state vector. The content generation unit is used to generate response feedback data, including error cause analysis and thought process guidance steps, as well as content retrieval instructions for retrieving teaching resources, based on the student interaction state vector.

[0014] Beneficial effects: The intent parsing unit generates a student interaction state vector containing knowledge point identifiers, knowledge point error probabilities, and response time classifications. The content generation unit then generates response feedback data based on this vector, achieving quantitative representation of learning cognitive states and automatic generation of customized feedback.

[0015] Furthermore, the student interaction state vector includes a knowledge point identifier field, a knowledge point error probability field, and a response time classification field.

[0016] Beneficial effects: By limiting the student interaction state vector to include knowledge point identifier fields, knowledge point error probability fields, and response time classification fields, a multi-dimensional structured expression of the student's cognitive state is achieved, which facilitates the calling and processing of subsequent modules.

[0017] Furthermore, the content generation unit uses an error classification function and a step decomposition function to generate response feedback data; the calculation logic of the error classification function and the step decomposition function are as follows: The calculation logic of the error classification function is as follows:

[0018] in, The feature vector output by the interactive intent recognition module. This serves as an identifier for the current knowledge point. For the preset set of error types, Given a feature vector and knowledge points Error types under conditions The conditional probability, The output is the incorrect classification result; The calculation logic of the step decomposition function is as follows: ,

[0019] in, To guide the sequence of steps, The total number of disassembly steps. For the first Step operation instruction text, For the first The validation node condition expression for the step. For the first The weighting coefficients of the step, the weighting coefficients Compared with the aforementioned misclassification results Associations are used to adjust the presentation order of the onboarding steps.

[0020] Beneficial effects: By generating response feedback data through error classification and step decomposition functions, calculating error types using conditional probability and adjusting the presentation order of guiding steps using weight coefficients, the accuracy of error identification and the personalization of feedback content are improved.

[0021] Furthermore, the teaching content storage module includes a structured database unit, an unstructured file storage unit, and a data update unit; The structured database unit uses a MySQL database to store structured teaching data; The unstructured file storage unit adopts a distributed file storage system for storing courseware, explanation videos, and knowledge point images; The data update unit is used to set up timed update tasks to update teaching resources and back up historical data within a preset time period.

[0022] Beneficial effects: By using a hierarchical storage architecture of structured database units and unstructured file storage units, combined with timed update units, the classification, storage and automatic updating of teaching resources are realized, improving the retrieval efficiency of teaching resources and data security.

[0023] Furthermore, the interactive module includes a student-side unit and a teacher-side unit; The student terminal unit includes a question answering interface, a question input interface, a feedback viewing interface, and a knowledge point review interface. The question input interface integrates a speech recognition API, which is used to convert speech input into text and transmit it to the data acquisition module. The teacher-side unit includes a student status monitoring interface, a feedback management interface, and a teaching adjustment interface. The student status monitoring interface is used to display students' answer accuracy rate, frequency of questions, and distribution of knowledge gaps in real time.

[0024] Beneficial effects: By setting up student-side and teacher-side units, the student-side integrates a speech recognition API to realize speech input conversion, while the teacher-side displays student learning status data in real time, thus achieving functional separation and data synchronization for two-way interaction between teachers and students. Attached Figure Description

[0025] This specification will be further described by way of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting; in these embodiments, the same reference numerals denote the same structures, wherein: Figure 1 This is an exemplary system architecture diagram of an interactive digital teaching system based on an AI big data model. Detailed Implementation

[0026] To more clearly illustrate the technical solutions of the embodiments in this specification, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are merely some examples or embodiments of this specification. For those skilled in the art, these drawings can be applied to other similar scenarios without creative effort. Unless obvious from the context or otherwise specified, the same reference numerals in the drawings represent the same structures or operations.

[0027] As indicated in this specification and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" do not specifically refer to the singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of expressly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0028] Flowcharts are used in this specification to illustrate the operations performed by the system according to embodiments of this specification. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, the steps can be processed in reverse order or simultaneously. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0029] The following detailed explanation illustrates the specific implementation methods: Example 1: like Figure 1 As shown, the interactive digital teaching system based on an AI big data model (hereinafter referred to as the teaching system) includes: The data acquisition module is used to collect real-time interactive data during student interactions, standardize the data, and send it to the AI ​​large model processing module.

[0030] Real-time interactive data refers to multi-dimensional behavioral data generated by students during their interaction with the teaching system. For example, real-time interactive data may include response text, response time, and operational behavior data.

[0031] Response text refers to the textual answers that students input in response to questions posed by the teaching system; response time refers to the total time a student spends from receiving a question to submitting a response, as well as the pause time during the answering process; and operation behavior data refers to the student's click operations and backtracking operation records during the interaction process.

[0032] In this embodiment, the data acquisition module specifically includes a text acquisition unit, a time acquisition unit, an operation acquisition unit, and a data integration unit.

[0033] The text acquisition unit is used to collect text data from the student's response content through the input interface of the interactive module, store it in UTF-8 encoding format, and remove invalid characters from the text data.

[0034] The time acquisition unit is used to collect the total time from when a student receives a question to when they submit their answer, as well as the pause time during the answering process, in real time through a timer to obtain response time data.

[0035] The threshold for collecting pause time can be set based on teaching experience, and the collection accuracy is accurate to milliseconds.

[0036] The operation acquisition unit is used to collect students' click operations and backtracking operations through the operation monitoring interface of the terminal device, record the operation type and operation time, and obtain behavioral operation data.

[0037] The terminal devices may include computers, tablets, and smartphones used by students.

[0038] The operation listening interface can be a native event listening interface provided by the teaching system or a behavior tracking interface embedded in the teaching client.

[0039] The data integration unit is used to integrate the data collected by the text acquisition unit, time acquisition unit and operation acquisition unit into JSON format data, and send it to the AI ​​large model processing module via HTTP protocol.

[0040] For example, when a student receives the question "Find the roots of the quadratic equation x²-5x+6=0" in a math class, the data acquisition module collects the student's input response text "x=2 and x=3", collects the total time from receiving the question to submitting the response (45 seconds), including an 8-second pause, and also collects the student's operation record of clicking the "View Hints" button during the answering process.

[0041] In this embodiment, the teaching system collects text data, response time data, and behavioral operation data through a text acquisition unit, a time acquisition unit, and an operation acquisition unit, respectively. The data is then integrated into JSON format by the data integration unit and sent via the HTTP protocol. This achieves structured acquisition and standardized transmission of multi-source interactive data, providing the AI ​​large model processing module with input data that is uniform in format and complete in features.

[0042] The interactive intent recognition module is used to extract feature vectors from student response texts and input them into the AI ​​large model processing module.

[0043] The interactive intent recognition module is built on the BERT model and includes a feature extraction unit and a feature filtering unit.

[0044] Among them, the BERT model is a pre-trained language representation model based on the Transformer architecture, which can perform deep semantic understanding of text and convert it into fixed-dimensional feature vectors.

[0045] The feature extraction unit is used to receive the text portion of the standardized real-time interactive data sent by the data acquisition module, extract semantic features, logical features, and error features from the student's response content, and output a feature vector.

[0046] Among them, semantic features refer to the core meaning expressed by the student's response; logical features refer to the reasoning chain and step relationship in the student's response; and error features refer to the specific location and type of deviation between the student's response and the correct answer.

[0047] In this embodiment, the feature extraction unit segments the text data into several tokens, adds positional encoding to each token, and inputs it into the pre-trained BERT model. The model performs bidirectional semantic encoding through a multi-layer Transformer encoder, extracts the 768-dimensional vector corresponding to the [CLS] token from the last hidden state as the overall semantic representation of the text data, and then maps this representation into the classification output of semantic features, logical features, and error features through a fully connected layer.

[0048] The feature filtering unit is used to remove invalid features that are irrelevant to the teaching content, retain the core features that are related to the knowledge points and questions, and send the filtered feature vectors to the AI ​​large model processing module.

[0049] In this embodiment, the feature filtering unit calculates the mutual information value between each dimension of the feature vector and the preset knowledge point label, determines the dimension with mutual information value lower than the preset threshold as invalid feature and sets it to zero, retains the dimension with mutual information value higher than the preset threshold as core feature, and adjusts the feature vector according to the question point identification result to generate the filtered feature vector and send it to the AI ​​large model processing module.

[0050] For example, continuing from the previous example, the interactive intent recognition module receives the student's response text "x=2 and x=3", extracts the semantic feature as "root result", the logical feature as "directly giving two numerical solutions", and the error feature as "not showing the solution process", and encodes the above features into a 768-dimensional feature vector and sends it to the AI ​​large model processing module.

[0051] In this embodiment, through the multi-layer processing of text data segmentation, encoding and mapping by the feature extraction unit, combined with the invalid feature elimination and weighted adjustment based on mutual information value by the feature filtering unit, the deep semantic, logical and error features in the student's response content are accurately extracted and irrelevant information is effectively filtered, thereby improving the representation quality of the feature vector and the efficiency of subsequent intent parsing.

[0052] The AI ​​large model processing module is used to receive standardized real-time interactive data and feature vectors, parse the real-time interactive data and feature vectors and determine the intent, and generate response feedback data and content retrieval instructions.

[0053] The AI ​​large-scale model processing module adopts a large language model based on the Transformer architecture, including an intent parsing unit and a content generation unit.

[0054] The intent parsing unit receives the feature vector sent by the interactive intent recognition module, and combines it with the response time data and behavioral operation data from the data acquisition module to generate a student interaction state vector.

[0055] The student interaction state vector is a multidimensional numerical vector used to represent the cognitive state of a student in the current interaction process.

[0056] In this embodiment, the student interaction state vector includes a knowledge point identifier field, a knowledge point error probability field, and a response time classification field.

[0057] In this embodiment, the knowledge point identifier field is used to identify the knowledge point involved in the current interaction; the knowledge point error probability field is used to quantify the likelihood of a student making a mistake on the current knowledge point; and the response time classification field is used to classify and judge the student's proficiency based on the time taken to answer the question. By integrating the information from the three dimensions of knowledge point identifier, error probability, and response time classification into a unified structured representation through the student interaction state vector, quantitative and directly callable student cognitive state data is provided for the subsequent content generation unit.

[0058] In this embodiment, the intent parsing unit performs multi-dimensional concatenation of feature vectors with response time data and behavioral operation data to form a fused feature matrix, which is then transformed nonlinearly by a fully connected neural network to output a fixed-dimensional student interaction state vector.

[0059] The content generation unit is used to generate response feedback data, including error cause analysis and thought process guidance steps, based on the student interaction state vector, as well as content retrieval instructions for accessing teaching resources.

[0060] The content generation unit uses an error classification function and a step decomposition function to generate response feedback data; the calculation logic of the error classification function and the step decomposition function are as follows: The calculation logic of the error classification function is as follows:

[0061] in, The feature vector output by the interactive intent recognition module. This serves as an identifier for the current knowledge point. For the preset set of error types, Given a feature vector and knowledge points Error types under conditions The conditional probability, This is the output of incorrect classification results.

[0062] The calculation logic of the step decomposition function is as follows: ,

[0063] in, To guide the sequence of steps, The total number of disassembly steps. For the first Step operation instruction text, For the first The validation node condition expression for the step. For the first The weighting coefficients of the step, the weighting coefficients Compared with the aforementioned misclassification results Associations are used to adjust the presentation order of the onboarding steps.

[0064] For example, continuing from the previous example, the AI ​​large model processing module identifies that although the student gave the correct result, they did not show the solution process, which is judged as a "process missing" error type. The generated response feedback data includes the error reason "the substitution steps of the quadratic equation root-finding formula were not shown" and the thought process guidance steps "Step 1: Determine the values ​​of a, b, and c in the equation, a=1, b=-5, c=6; Step 2: Calculate the discriminant Δ=b²-4ac=25-24=1; Step 3: Substitute into the root-finding formula x=(-b±√Δ) / 2a, and calculate x=2 and x=3". At the same time, a content retrieval command is generated to retrieve relevant teaching resources on "derivation of the quadratic equation root-finding formula".

[0065] The teaching content storage module is used to store teaching resources and, according to the content retrieval instructions output by the AI ​​large model processing module, retrieve the corresponding teaching resources and push them to the interactive module.

[0066] The teaching content storage module includes a structured database unit, an unstructured file storage unit, and a data update unit.

[0067] The structured database unit uses a MySQL database to store structured teaching data.

[0068] The unstructured file storage unit uses a distributed file storage system to store courseware, explanation videos, and images of key knowledge points. Each file corresponds to a unique index value.

[0069] The data update unit is used to set up scheduled update tasks, update teaching resources and back up historical data within a preset time period.

[0070] The interactive module is used to receive responses input by students; to display teaching resources pushed by the teaching content storage module and response feedback data generated by the AI ​​large model processing module; and to synchronize the responses input by students to the data acquisition module.

[0071] In this embodiment, the interactive module includes a student-side unit and a teacher-side unit.

[0072] The student-side unit includes a question-answering interface, a question input interface, a feedback viewing interface, and a knowledge point review interface. The question input interface integrates a speech recognition API, which is used to convert speech input into text and transmit it to the data acquisition module.

[0073] The question-answering interface supports interactive methods such as text input and option selection; the question input interface integrates a speech recognition API to convert speech input into text and transmit it to the data acquisition module; the feedback viewing interface displays personalized response feedback data generated by the AI ​​large model processing module in real time; and the knowledge point review interface displays targeted review content pushed by the system.

[0074] The teacher-side unit includes a student status monitoring interface, a feedback management interface, and a teaching adjustment interface. The student status monitoring interface is used to display students' answer accuracy rate, frequency of questions, and distribution of knowledge gaps in real time.

[0075] For example, after a student submits their answer through the student-side unit's answer interface, the interactive module displays the error analysis and thought process guidance steps generated by the AI ​​large model processing module, while also showing the "Derivation of the Formula for Solving Quadratic Equations" video courseware pushed by the teaching content storage module. The teacher can view the student's answer accuracy, frequency of questions, and distribution of knowledge gaps through the student status monitoring interface of the teacher-side unit, and focus on the student's mastery of the "quadratic equations" knowledge point based on the feedback data generated by the system.

[0076] In this embodiment, by setting up a student-side unit that includes a question-answering interface, a question input interface, a feedback viewing interface, and a knowledge point review interface, and a teacher-side unit that includes a student status monitoring interface, a feedback management interface, and a teaching adjustment interface, the functional separation and data synchronization of two-way interaction between teachers and students are realized. The integration of a speech recognition API on the student-side reduces the threshold for text input, and the real-time display of student learning status data on the teacher-side facilitates timely intervention, thereby improving the convenience of teaching interaction and the pertinence of teaching management.

[0077] The basic concepts have been described above. Obviously, for those skilled in the art, the detailed disclosure above is merely illustrative and does not constitute a limitation of this specification. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements, and corrections to this specification. Such modifications, improvements, and corrections are suggested in this specification and therefore remain within the spirit and scope of the exemplary embodiments described herein.

[0078] Furthermore, unless expressly stated in the claims, the order of processing elements and sequences, the use of numbers and letters, or other names described in this specification are not intended to limit the order of the processes and methods described herein. Although various examples have been discussed in the foregoing disclosure of some embodiments of the invention that are currently considered useful, it should be understood that such details are for illustrative purposes only, and the appended claims are not limited to the disclosed embodiments; rather, the claims are intended to cover all modifications and equivalent combinations that conform to the spirit and scope of the embodiments described herein. For example, while the system components described above can be implemented using hardware devices, they can also be implemented solely using software solutions, such as installing the described system on existing servers or mobile devices.

[0079] Similarly, it should be noted that, in order to simplify the description disclosed herein and thus aid in the understanding of one or more embodiments of the invention, the foregoing description of embodiments in this specification may sometimes combine multiple features into a single embodiment, drawing, or description thereof. However, this method of disclosure does not imply that the subject matter of this specification requires more features than those mentioned in the claims. In fact, the embodiments contain fewer features than all the features of a single embodiment disclosed above.

[0080] In some embodiments, numbers describing the quantity of components and attributes are used. It should be understood that such numbers used in the description of embodiments are modified in some examples with the terms "approximately," "approximately," or "generally." Unless otherwise stated, "approximately," "approximately," or "generally" indicates that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which may be changed depending on the characteristics required by individual embodiments. In some embodiments, numerical parameters should take into account specified significant digits and employ a general method of digit reservation. Although the numerical ranges and parameters used to confirm their breadth of range in some embodiments of this specification are approximate values, in specific embodiments, such values ​​are set as precisely as feasible.

[0081] For each patent, patent application, patent application publication, and other material such as articles, books, specifications, publications, and documents referenced in this specification, the entire contents of which are incorporated herein by reference. This excludes historical application documents that are inconsistent with or conflict with the content of this specification, as well as documents that limit the broadest scope of the claims in this specification (currently or subsequently appended to this specification). It should be noted that in the event of any inconsistency or conflict between the descriptions, definitions, and / or terminology used in the supplementary materials to this specification and the content of this specification, the descriptions, definitions, and / or terminology used in this specification shall prevail.

[0082] Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments described herein. Other variations may also fall within the scope of this specification. Therefore, alternative configurations of the embodiments described herein are intended to be illustrative rather than limiting, and should be considered consistent with the teachings of this specification. Accordingly, the embodiments described herein are not limited to those explicitly introduced and described herein.

Claims

1. An interactive digital teaching system based on an AI large-scale model, characterized in that: include: The data acquisition module is used to collect real-time interactive data during student interaction, perform standardized processing, and send it to the AI ​​large model processing module. The interactive intent recognition module is used to extract feature vectors from student response texts and input them into the AI ​​large model processing module; The AI ​​large model processing module is used to receive standardized real-time interactive data and feature vectors, parse the real-time interactive data and feature vectors and determine the intent, and generate response feedback data and content retrieval instructions. The teaching content storage module is used to store teaching resources and, according to the content retrieval instructions output by the AI ​​large model processing module, retrieve the corresponding teaching resources and push them to the interactive module. The interactive module is used to receive the response content input by the student; to display the teaching resources pushed by the teaching content storage module and the response feedback data generated by the AI ​​big model processing module to the student; and to synchronize the response content input by the student to the data acquisition module.

2. The interactive digital teaching system based on an AI large-scale model according to claim 1, characterized in that, The data acquisition module includes a text acquisition unit, a time acquisition unit, an operation acquisition unit, and a data integration unit; The text acquisition unit is used to acquire text data from the student's response content through the input interface of the interactive module, store it in UTF-8 encoding format, and remove invalid characters from the text data. The time acquisition unit is used to collect the total time from when the student receives the question to when they submit their answer, as well as the pause time during the answering process, in real time through a timer to obtain response time data; The operation acquisition unit is used to collect students' click operations and backtracking operations through the operation monitoring interface of the terminal device, record the operation type and operation time, and obtain behavioral operation data. The data integration unit is used to integrate the data collected by the text acquisition unit, the time acquisition unit and the operation acquisition unit into JSON format data, and send it to the AI ​​large model processing module via HTTP protocol.

3. The interactive digital teaching system based on an AI large-scale model according to claim 1, characterized in that, The interactive intent recognition module is built based on the BERT model and includes a feature extraction unit and a feature filtering unit; The feature extraction unit is used to receive the text portion of the standardized real-time interactive data sent by the data acquisition module, extract the semantic features, logical features, and error features from the student's response content, and output a feature vector. The feature filtering unit is used to remove invalid features that are irrelevant to the teaching content, retain core features that are related to knowledge points and questions, and send the filtered feature vectors to the AI ​​large model processing module.

4. The interactive digital teaching system based on an AI large model according to claim 1, characterized in that, The AI ​​large model processing module adopts a large language model based on the Transformer architecture, including an intent parsing unit and a content generation unit; The intent parsing unit is used to receive the feature vector sent by the interactive intent recognition module, and combine it with the response time data and behavioral operation data of the data acquisition module to generate a student interaction state vector. The content generation unit is used to generate response feedback data, including error cause analysis and thought process guidance steps, as well as content retrieval instructions for retrieving teaching resources, based on the student interaction state vector.

5. The interactive digital teaching system based on an AI large model according to claim 4, characterized in that, The student interaction state vector includes a knowledge point identifier field, a knowledge point error probability field, and a response time classification field.

6. The interactive digital teaching system based on an AI large model according to claim 4, characterized in that, The content generation unit uses an error classification function and a step decomposition function to generate response feedback data; the calculation logic of the error classification function and the step decomposition function are as follows: The calculation logic of the error classification function is as follows: in, The feature vector output by the interactive intent recognition module. This serves as an identifier for the current knowledge point. For the preset set of error types, Given a feature vector and knowledge points Error types under conditions The conditional probability, The output is the incorrect classification result; The calculation logic of the step decomposition function is as follows: ， in, To guide the sequence of steps, The total number of disassembly steps. For the first Step operation instruction text, For the first The validation node condition expression for the step. For the first The weighting coefficients of the step, the weighting coefficients Compared with the aforementioned misclassification results Associations are used to adjust the presentation order of the onboarding steps.

7. The interactive digital teaching system based on an AI large model according to claim 1, characterized in that, The teaching content storage module includes a structured database unit, an unstructured file storage unit, and a data update unit; The structured database unit uses a MySQL database to store structured teaching data; The unstructured file storage unit adopts a distributed file storage system for storing courseware, explanation videos, and knowledge point images; The data update unit is used to set up timed update tasks to update teaching resources and back up historical data within a preset time period.

8. The interactive digital teaching system based on an AI large model according to claim 1, characterized in that, The interactive module includes a student terminal and a teacher terminal; The student terminal unit includes a question answering interface, a question input interface, a feedback viewing interface, and a knowledge point review interface. The question input interface integrates a speech recognition API, which is used to convert speech input into text and transmit it to the data acquisition module. The teacher-side unit includes a student status monitoring interface, a feedback management interface, and a teaching adjustment interface. The student status monitoring interface is used to display students' answer accuracy rate, frequency of questions, and distribution of knowledge gaps in real time.

Images