LLM-based multi-source education information recommendation method and device

Abstract

The invention discloses an LLM-based multi-source education information recommendation method and device, and relates to the technical field of intelligent education, and the method comprises the steps: separately collecting education information, subject knowledge information and subject test question information, carrying out text vectorization by using document slices, and storing the vectorized information into a vector database to obtain a multi-source education information database; when a user initiates a request related to subject knowledge and education information, request content is obtained, intention analysis is performed through LLM, and target information corresponding to the request is retrieved from the multi-source education information database according to an intention analysis result; and when the request of the user has a demand for searching the related test questions, initiating a retrieval demand to the Agent tool responsible for test question retrieval through the LLM, and obtaining the target test questions from the multi-source education information database by utilizing the Agent tool. The problem that an education platform in the prior art cannot meet education multi-scene requirements and accurate recommendation is solved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent education technology, and in particular to a method and device for recommending multi-source education information based on LLM. Background Art

[0002] K12 education has a series of scenario requirements, such as information query, knowledge point query, test question acquisition, test question evaluation and ability evaluation, etc. The subject education platforms on the market are basically data display platforms, which cannot meet the above-mentioned multiple scenario requirements, and the search is still at the stage of keyword matching, which cannot achieve accurate matching of user needs. Summary of the invention

[0003] In view of this, the purpose of the present invention is to provide a method and device for recommending multi-source educational information based on LLM, aiming to solve the problem that educational platforms in the prior art cannot meet the needs of multiple educational scenarios and accurate recommendations.

[0004] In one aspect, the present invention provides a method for recommending multi-source educational information based on LLM, the method comprising: Collecting educational information, subject knowledge information and subject test information respectively, performing text vectorization on the educational information, subject knowledge information and subject test information respectively by using document slicing, and then storing them in a vector database to obtain a multi-source educational information database; When a user initiates a request related to subject knowledge or educational information, the request content is obtained and analyzed through LLM. Based on the intention analysis results, the target information corresponding to the request is retrieved from the multi-source educational information database. When the user's request contains the need to find relevant test questions, the search request is initiated to the Agent tool responsible for test question retrieval through LLM, and the Agent tool is used to obtain the target test questions from the multi-source education information database.

[0005] Furthermore, in the above-mentioned method for recommending multi-source educational information, the step of using document slicing to perform text vectorization includes: Preprocessing the acquired documents, wherein the preprocessing includes denoising, standardization, and paragraph segmentation; Use the preset syntax analysis algorithm to perform semantic analysis on the paragraph content, and slice the paragraph content according to the analysis results; Perform structural analysis on the sliced ​​content and slide analysis using the context window, and then adjust the slice content to ensure that the length of each slice content is within the preset range; The slice content is input into the pre-trained language model for vectorization to perform text vectorization on the slice content.

[0006] Furthermore, in the above-mentioned method for recommending multi-source educational information, the steps of obtaining request content and performing intent analysis through LLM include: Obtain the language meaning and context information of the request content to identify the instruction type and specific field of the request content; Identify the semantic information of the request content, and determine the intent of the request content by combining the instruction type and specific domain of the request content.

[0007] Furthermore, in the above-mentioned method for recommending multi-source educational information, the step of retrieving target information corresponding to the request from the multi-source educational information database according to the intention analysis result comprises: Obtaining the user's portrait feature information, and retrieving the target information corresponding to the request from the multi-source education information database according to the portrait feature information and the intention analysis result; Among them, by analyzing the user's questions about subject knowledge, combined with the user's grade, the overall mastery level of the target information by students in that grade is determined, thereby comparing the subject level of the user in the corresponding grade stage; Obtain the user's test results for relevant test questions, determine the user's target subject level based on the test results and subject level, and thus determine the user's portrait feature information.

[0008] Furthermore, the method for recommending multi-source educational information further comprises: When the user's request is for information query, relevant information or news is retrieved from the multi-source education information database and fed back to the user according to the user's query subject; When the user's request is for consulting opinions, the topics involved in the consulting opinions are retrieved from the multi-source education information database, summarized and refined, and then fed back to the user; When the user's request is for creative tips, creative suggestions are given based on the relevant information corresponding to the request.

[0009] Furthermore, in the above-mentioned method for recommending multi-source educational information, the steps of respectively collecting educational information, subject knowledge information and subject test information include: Education information is captured from the portal website in real time through the crawler system, and subject knowledge information and subject test question information are collected from the preset subject knowledge database and subject test question database respectively.

[0010] Another object of the present invention is to provide a device for recommending multi-source educational information based on LLM, the device comprising: An acquisition module is used to collect educational information, subject knowledge information and subject test information respectively, and to vectorize the educational information, subject knowledge information and subject test information respectively by using document slicing, and then store them in a vector database to obtain a multi-source educational information database; The first retrieval module is used to obtain the request content and perform intent analysis through LLM when the user initiates a request related to subject knowledge and educational information, and retrieve the target information corresponding to the request from the multi-source educational information database based on the intent analysis result; The second retrieval module is used to initiate a retrieval request to the Agent tool responsible for question retrieval through LLM when the user's request requires the search for relevant questions, and use the Agent tool to obtain target questions from a multi-source education information database.

[0011] Another object of the present invention is to provide a readable storage medium having a computer program stored thereon, wherein the program implements the steps of the above method when executed by a processor.

[0012] Another object of the present invention is to provide an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the program.

[0013] The present invention collects educational information, subject knowledge information and subject test information respectively, and uses document slicing to vectorize the text of educational information, subject knowledge information and subject test information respectively, and then stores them in a vector database to obtain a multi-source educational information database, which integrates the information required for various scenarios in education; when a user initiates a request related to subject knowledge and educational information, the request content is obtained and intent analysis is performed through LLM, and the target information corresponding to the request is retrieved from the multi-source educational information database according to the intent analysis result; when the user's request has a need to find related test questions, a search demand is initiated to the Agent tool responsible for test question retrieval through LLM, and the target test questions are obtained from the multi-source educational information database using the Agent tool, and according to different actual needs, intent analysis is performed through LLM, and personalized and accurate recommendations are given. This solves the problem that the education platform in the prior art cannot meet the needs of multiple education scenarios and accurate recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a flowchart of a method for recommending multi-source educational information based on LLM in a first embodiment of the present invention; Figure 2 It is a structural block diagram of a device for recommending multi-source educational information based on LLM in the third embodiment of the present invention.

[0015] The following specific implementation manner will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0016] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0017] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0019] The following will explain in detail how to meet the current multi-scenario needs of education and accurately recommend educational information in combination with specific embodiments and drawings.

[0020] Embodiment 1 See also Figure 1 , shown is a method for recommending multi-source educational information based on LLM in the first embodiment of the present invention, and the method includes steps S10 to S12.

[0021] Step S10, respectively collect educational information, subject knowledge information and subject test information, respectively perform text vectorization on the educational information, subject knowledge information and subject test information using document slicing, and then store them in a vector database to obtain a multi-source educational information database.

[0022] Specifically, the embodiment of the present invention is implemented based on an integrated education platform, which integrates an education information section, a subject knowledge section and a subject test section. Through the planning capability of LLM, multi-agent tools are called to integrate real-time education information, subject knowledge, subject test recommendation and other services into one platform, providing users with an interactive comprehensive education service for all scenarios of subject education.

[0023] Among them, the platform uses a crawler system to capture the latest education information from mainstream portals in real time. The acquired education information is vectorized through document slicing and stored in a vector database to obtain a multi-source education information database. At the same time, a document slice index is established to realize vectorized retrieval of information information. The platform is also linked to the preset subject knowledge database and the preset subject test question database. The subject knowledge mainly includes the textbook knowledge system of various subjects in the K12 education stage. Subject knowledge and subject test questions will also be vectorized through document slicing to meet vectorized retrieval needs.

[0024] Furthermore, document slicing must ensure the coherence of content and the consistency of the subject. The amount of information in the document slice should be moderate, while meeting the retrieval requirements and ensuring the integrity of the title and content as much as possible. It is also necessary to follow a reasonable granularity division. When slicing documents, the amount of text should be smaller while ensuring semantic integrity. Exemplarily, the acquired documents may be preprocessed, wherein the preprocessing includes denoising, standardization, and paragraph segmentation; semantic analysis is performed on the paragraph content using a preset syntactic analysis algorithm; the paragraph content is sliced ​​according to the analysis results to avoid segmenting closely related sentences into different segments; and if a paragraph contains a complex sentence structure associated with the context, it should be segmented as a whole paragraph to avoid destroying the semantic coherence; structural analysis is performed on the sliced ​​content and sliding analysis is performed using a context window, and then the slice content is adjusted to ensure that the length of each slice content is within a preset range, and the amount of information in the slice needs to be moderate, neither too short (information fragmentation will affect the retrieval effect) nor too long (search efficiency will be reduced after vectorization); finally, the slice content is input into a pre-trained language model (such as BERT, RoBERTa, etc.) for vectorization to perform text vectorization on the slice content.

[0025] Step S11, when a user initiates a request related to subject knowledge or educational information, the request content is obtained and intent analysis is performed through LLM, and the target information corresponding to the request is retrieved from the multi-source educational information database based on the intent analysis result.

[0026] Among them, the Large Language Model (LLM) has the ability to analyze intent. When receiving requests related to subject knowledge and educational information, it analyzes the user's intent and then retrieves the target information required by the relevant user from the multi-source education information database.

[0027] Specifically, the big language model will classify the intent of the question based on the linguistic meaning and contextual information of the request content, usually including identifying the type of question (such as inquiry, request, command, etc.) and specific field (such as technology, education, entertainment, etc.). By integrating this information, the big model can form a comprehensive understanding of the user's true intent and answer the user's questions more accurately.

[0028] Step S12: When the user's request contains a need to search for relevant test questions, a search request is initiated to the Agent tool responsible for test question search through the LLM, and the Agent tool is used to obtain target test questions from a multi-source education information database.

[0029] Among them, if the user's request contains the need to find relevant test questions, then LLM will initiate a search request to the Agent tool responsible for test question retrieval, obtain relevant test questions and feedback them to the user.

[0030] In summary, the method for recommending multi-source educational information based on LLM in the above embodiment of the present invention collects educational information, subject knowledge information and subject test information respectively, and uses document slicing to perform text vectorization on the educational information, subject knowledge information and subject test information respectively, and then stores them in the vector database to obtain a multi-source educational information database, which integrates the information required for various scenarios in education; when a user initiates a request related to subject knowledge and educational information, the request content is obtained and intent analysis is performed through LLM, and the target information corresponding to the request is retrieved from the multi-source educational information database according to the intent analysis result; when the user's request contains a need to find relevant test questions, a search demand is initiated to the Agent tool responsible for test question retrieval through LLM, and the target test questions are obtained from the multi-source educational information database using the Agent tool, and according to different actual needs, intent analysis is performed through LLM, and personalized and accurate recommendations are given. This solves the problem that the educational platform in the prior art cannot meet the needs of multiple educational scenarios and accurate recommendations.

[0031] Embodiment 2 This embodiment also proposes a method for recommending multi-source educational information based on LLM. The difference between the method for recommending multi-source educational information based on LLM in this embodiment and the method for recommending multi-source educational information based on LLM in the first embodiment is that: The step of retrieving target information corresponding to the request from the multi-source education information database according to the intention analysis result comprises: Obtaining the user's portrait feature information, and retrieving the target information corresponding to the request from the multi-source education information database according to the portrait feature information and the intention analysis result; Among them, by analyzing the user's questions about subject knowledge, combined with the user's grade, the overall mastery level of the target information by students in that grade is determined, thereby comparing the subject level of the user in the corresponding grade stage; Obtain the user's test results for relevant test questions, determine the user's target subject level based on the test results and subject level, and thus determine the user's portrait feature information.

[0032] Among them, LLM is combined with user query intent and user portrait feature information to achieve accurate information push. Specifically, by analyzing the user's questions about subject knowledge and combining the user's personal information (grade, subject, etc.), through background big data analysis, we can know the overall mastery level of the target information by students in that grade, and thus compare the subject level of the user in the grade. Then, through the test results of a series of related questions, further analysis and summary are conducted, and the user's subject level positioning judgment is optimized, so as to provide a dynamic and scientific user portrait feature information.

[0033] In addition, in some optional embodiments of the present invention, the method further includes: When the user's request is for information query, relevant information or news is retrieved from the multi-source education information database and fed back to the user according to the user's query subject; When the user's request is for consulting opinions, the topics involved in the consulting opinions are retrieved from the multi-source education information database, summarized and refined, and then fed back to the user; When the user's request is for creative tips, creative suggestions are given based on the relevant information corresponding to the request.

[0034] Among them, when the user raises a relevant question request, LLM first analyzes the request information to learn the specific purpose of the user's request (query, consultation or creative prompt, etc.), and then gives corresponding replies according to the different purposes of the user. If the user's purpose is to query information, then the database will be retrieved for relevant information or news based on the user's query subject and feedback will be given to the user. If the user's purpose is to consult, LLM will further summarize and refine the relevant information in the database based on the subject involved in the user's question and feedback to the user. If the user's purpose is to give creative prompts, then LLM will further diverge based on the relevant information, use LLM's generative potential, and give reasonable creative suggestions.

[0035] In summary, the method for recommending multi-source educational information based on LLM in the above embodiment of the present invention collects educational information, subject knowledge information and subject test information respectively, and uses document slicing to perform text vectorization on the educational information, subject knowledge information and subject test information respectively, and then stores them in the vector database to obtain a multi-source educational information database, which integrates the information required for various scenarios in education; when a user initiates a request related to subject knowledge and educational information, the request content is obtained and intent analysis is performed through LLM, and the target information corresponding to the request is retrieved from the multi-source educational information database according to the intent analysis result; when the user's request contains a need to find relevant test questions, a search demand is initiated to the Agent tool responsible for test question retrieval through LLM, and the target test questions are obtained from the multi-source educational information database using the Agent tool, and according to different actual needs, intent analysis is performed through LLM, and personalized and accurate recommendations are given. This solves the problem that the educational platform in the prior art cannot meet the needs of multiple educational scenarios and accurate recommendations.

[0036] Embodiment 3 See also Figure 2 , shown is a device for recommending multi-source educational information based on LLM proposed in the third embodiment of the present invention, the device comprising: The acquisition module 100 is used to respectively collect educational information, subject knowledge information and subject test information, respectively perform text vectorization on the educational information, subject knowledge information and subject test information using document slicing, and then store them in a vector database to obtain a multi-source educational information database; The first search module 200 is used to obtain the request content and perform intent analysis through LLM when the user initiates a request related to subject knowledge and educational information, and retrieve the target information corresponding to the request from the multi-source educational information database according to the intent analysis result; The second search module 300 is used to initiate a search request to the Agent tool responsible for question search through LLM when the user's request has a need to search for relevant questions, and use the Agent tool to obtain target questions from a multi-source education information database.

[0037] Furthermore, in the above-mentioned device for recommending multi-source educational information, the step of using document slicing to perform text vectorization includes: Preprocessing the acquired documents, wherein the preprocessing includes denoising, standardization, and paragraph segmentation; Use the preset syntax analysis algorithm to perform semantic analysis on the paragraph content, and slice the paragraph content according to the analysis results; Perform structural analysis on the sliced ​​content and slide analysis using the context window, and then adjust the slice content to ensure that the length of each slice content is within the preset range; The slice content is input into the pre-trained language model for vectorization to perform text vectorization on the slice content.

[0038] Furthermore, in the above-mentioned device for recommending multi-source educational information, the step of obtaining the request content and performing intent analysis through LLM comprises: Obtain the language meaning and context information of the request content to identify the instruction type and specific field of the request content; Identify the semantic information of the request content, and determine the intent of the request content by combining the instruction type and specific domain of the request content.

[0039] Furthermore, in the above-mentioned device for recommending multi-source educational information, the step of retrieving target information corresponding to the request from the multi-source educational information database according to the intention analysis result comprises: Obtaining the user's portrait feature information, and retrieving the target information corresponding to the request from the multi-source education information database according to the portrait feature information and the intention analysis result; Among them, by analyzing the user's questions about subject knowledge, combined with the user's grade, the overall mastery level of the target information by students in that grade is determined, thereby comparing the subject level of the user in the corresponding grade stage; Obtain the user's test results for relevant test questions, determine the user's target subject level based on the test results and subject level, and thus determine the user's portrait feature information.

[0040] Furthermore, the device for recommending multi-source educational information further comprises: When the user's request is for information query, relevant information or news is retrieved from the multi-source education information database and fed back to the user according to the user's query subject; When the user's request is for consulting opinions, the topics involved in the consulting opinions are retrieved from the multi-source education information database, summarized and refined, and then fed back to the user; When the user's request is for creative tips, creative suggestions are given based on the relevant information corresponding to the request.

[0041] Furthermore, in the above-mentioned device for recommending multi-source educational information, the steps of respectively collecting educational information, subject knowledge information and subject test information include: Education information is captured from the portal website in real time through the crawler system, and subject knowledge information and subject test question information are collected from the preset subject knowledge database and subject test question database respectively.

[0042] The functions or operation steps implemented when the above modules are executed are substantially the same as those in the above method embodiments, and will not be repeated here.

[0043] Embodiment 4 Another aspect of the present invention further provides a readable storage medium having a computer program stored thereon, and when the program is executed by a processor, the steps of the method described in any one of the above-mentioned embodiments 1 to 2 are implemented.

[0044] Embodiment 5 Another aspect of the present invention provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method described in any one of the above-mentioned embodiments 1 to 2 when executing the program.

[0045] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0046] Those skilled in the art will appreciate that the logic and / or steps represented in the flowchart or otherwise described herein, for example, may be considered as an ordered list of executable instructions for implementing logical functions, and may be specifically implemented in any computer-readable storage medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For purposes of this specification, a "computer-readable storage medium" may be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0047] More specific examples (a non-exhaustive list) of computer-readable storage media include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable storage medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0048] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or a combination thereof: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0049] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0050] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A method for recommending multi-source educational information based on LLM, characterized in that: The method comprises: Collecting educational information, subject knowledge information and subject test information respectively, performing text vectorization on the educational information, subject knowledge information and subject test information respectively by using document slicing, and then storing them in a vector database to obtain a multi-source educational information database; When a user initiates a request related to subject knowledge or educational information, the request content is obtained and analyzed through LLM. Based on the intention analysis results, the target information corresponding to the request is retrieved from the multi-source educational information database. When the user's request contains the need to search for relevant test questions, the LLM initiates a search request to the Agent tool responsible for test question retrieval, and the Agent tool is used to obtain the target test questions from the multi-source education information database; The step of using document slicing to perform text vectorization includes: Preprocessing the acquired documents, wherein the preprocessing includes denoising, standardization, and paragraph segmentation; Use the preset syntax analysis algorithm to perform semantic analysis on the paragraph content, and slice the paragraph content according to the analysis results; Perform structural analysis on the sliced ​​content and slide analysis using the context window, and then adjust the slice content to ensure that the length of each slice content is within the preset range; The slice content is input into the pre-trained language model for vectorization to perform text vectorization on the slice content.

2. The method for recommending multi-source educational information based on LLM according to claim 1, characterized in that: The steps of obtaining the request content and performing intent analysis through LLM include: Obtain the language meaning and context information of the request content to identify the instruction type and specific field of the request content; Identify the semantic information of the request content, and determine the intent of the request content by combining the instruction type and specific domain of the request content.

3. The method for recommending multi-source educational information based on LLM according to claim 1, characterized in that: The step of retrieving target information corresponding to the request from the multi-source education information database according to the intention analysis result comprises: Obtaining the user's portrait feature information, and retrieving the target information corresponding to the request from the multi-source education information database according to the portrait feature information and the intention analysis result; Among them, by analyzing the user's questions about subject knowledge, combined with the user's grade, the overall mastery level of the target information by students in that grade is determined, thereby comparing the subject level of the user in the corresponding grade stage; Obtain the user's test results for relevant test questions, determine the user's target subject level based on the test results and subject level, and thus determine the user's portrait feature information.

4. The method for recommending multi-source educational information based on LLM according to claim 1, characterized in that: The method further comprises: When the user's request is for information query, relevant information or news is retrieved from the multi-source education information database and fed back to the user according to the user's query subject; When the user's request is for consulting opinions, the topics involved in the consulting opinions are retrieved from the multi-source education information database, summarized and refined, and then fed back to the user; When the user's request is for creative tips, creative suggestions are given based on the relevant information corresponding to the request.

5. The method for recommending multi-source educational information based on LLM according to claim 1, characterized in that: The steps of respectively collecting educational information, subject knowledge information and subject test information include: The educational information is captured in real time from the portal website through the crawler system, and the subject knowledge information and subject test question information are collected from the preset subject knowledge database and subject test question database respectively.

6. A device for recommending multi-source educational information based on LLM, characterized in that: The device is used to implement the method for recommending multi-source educational information based on LLM as described in any one of claims 1 to 5, comprising: An acquisition module is used to collect educational information, subject knowledge information and subject test information respectively, and to vectorize the educational information, subject knowledge information and subject test information respectively by using document slicing, and then store them in a vector database to obtain a multi-source educational information database; The first retrieval module is used to obtain the request content and perform intent analysis through LLM when the user initiates a request related to subject knowledge and educational information, and retrieve the target information corresponding to the request from the multi-source educational information database based on the intent analysis result; The second retrieval module is used to initiate a retrieval request to the Agent tool responsible for question retrieval through LLM when the user's request requires the search for relevant questions, and use the Agent tool to obtain target questions from a multi-source education information database.

7. A readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

8. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method according to any one of claims 1 to 5 when executing the program.

Images