Retrieval enhancement generation method and device based on knowledge graph in medical education scene

Abstract

The invention discloses an enhanced generation method and device based on knowledge graph retrieval in a medical education scene. The method comprises the following steps: segmenting a medical textbook; attaching the segmented text segments to an existing medical education knowledge graph; updating the attributes of the nodes and the edges by using a large language model technology; performing embedded representation on the updated medical education knowledge graph; performing embedded representation on the segmented text segments through a vectorization model; performing entity identification on the query of the user, and recalling related node vectors to obtain a total query node vector set; screening out a literature set; screening a document set from the text vector database; and according to the query of the user, the total node vector set, the literature set and the document set, generating an answer through a large language model and returning the answer to the user. According to the method, the medical education knowledge base can be effectively constructed and updated, and high-quality and accurate answers can be provided for questions in medical education.

Description

Technical Field

[0001] The present invention belongs to the field of medical natural language processing technology, and in particular relates to a knowledge graph retrieval enhancement generation method and device in a medical education scenario. Background Art

[0002] With the rapid development of artificial intelligence and data processing technologies, knowledge graphs and natural language processing (NLP) are increasingly being used in various fields. In medical education, accurately and rapidly acquiring and processing complex medical knowledge is crucial for improving educational quality and scientific research efficiency. Traditional medical education relies primarily on textbooks, literature, and the accumulation of clinical experience. However, this knowledge is fragmented and difficult to systematically manage and retrieve. This makes it particularly difficult to provide high-quality and accurate answers to open questions and answers in medical education.

[0003] As an emerging knowledge representation method, knowledge graphs, by storing knowledge in a structured graph format, not only effectively display relationships between entities but also enable efficient knowledge retrieval and reasoning through graph embedding techniques. In recent years, the development of large language models (LLMs), such as GPT-3 and BERT, has demonstrated powerful capabilities in processing natural language tasks. These models can understand and generate natural language text, perform semantic analysis, and perform complex language reasoning, providing strong support for the construction and updating of knowledge graphs.

[0004] In medical education scenarios, the combination of large language models and knowledge graphs can effectively address the information silos and inefficient retrieval issues inherent in traditional methods. By semantically segmenting medical textbooks and literature and attaching them to a medical education knowledge graph, systematic management of medical knowledge can be achieved. Furthermore, the use of large language model technology allows for dynamic updating of node and edge attributes within the knowledge graph, enhancing its expressiveness and accuracy. Applying graph embedding algorithms to embed the knowledge graph, generate a set of node vectors, and store them in a vector database can significantly improve the speed and accuracy of knowledge retrieval.

[0005] For user queries, entity recognition can be used to retrieve relevant nodes in the knowledge graph. Similarity calculation methods can then be used to filter out highly relevant text from the associated literature and textbook text vector library. Finally, a large language model is used to integrate these search results, generating accurate, complete, and logically rigorous answers that are returned to the user. This approach not only improves the efficiency and accuracy of knowledge retrieval in medical education, but also provides high-quality open question-and-answer services, significantly improving the teaching and research levels of medical education.

[0006] Therefore, the open question answering method based on knowledge graph retrieval enhancement generation in medical education scenarios has both academic research value and broad application prospects. It not only provides new ideas for knowledge management and retrieval in medical education, but also provides a useful reference for the application of knowledge graphs in other fields. Summary of the Invention

[0007] The purpose of the present invention is to address the deficiencies of the existing technology and provide a method and device for enhanced generation based on knowledge graph retrieval in a medical education scenario.

[0008] The object of the present invention is achieved through the following technical solution: a knowledge graph retrieval enhancement generation method based on medical education scenario, comprising the following steps: (1) Use the document segmentation algorithm to semantically segment the text content of the medical textbook, ensuring that each segmented part is semantically independent and complete, and obtain N segmented text fragments: ,in, Represents any segmented text fragment, ; (2) attaching the N segmented text segments to the existing medical education knowledge graph to obtain the attached medical education knowledge graph; (3) Use the large language model technology to update the attributes of the nodes and edges in the affiliated medical education knowledge graph to obtain the updated medical education knowledge graph; (4) Applying the graph embedding algorithm, embedding the updated medical education knowledge graph, generating a node vector set, and storing it in the graph vector database; (5) Each of the N segmented text segments is embedded and represented by a vectorization model, and a document vector set corresponding to each segmented text segment is generated and stored in a text vector database; (6) Perform entity recognition on the user's query and recall relevant node vectors in the graph vector database to obtain the total set of query node vectors corresponding to the user's query; (7) Using the text similarity calculation method, a collection of documents related to the user's query is screened out from the text fragments corresponding to each node vector in the total set of query node vectors; (8) Using the text similarity calculation method, we can directly filter out the document set related to the user's query in the text vector database; (9) Based on the user's query, the total set of query node vectors, the document set, and the document set, the large language model generates the answer corresponding to the query and returns it to the user.

[0009] Furthermore, the step (2) specifically includes the following sub-steps: (2.1) Traverse each text segment after segmentation , perform named entity recognition and extract text fragments Medical entities in form medical entity sets : ,in, Represents a text fragment Extractable medical entities, Represents a text fragment The extractable medical entities; (2.2) Medical entity collection Each medical entity The entity linking technology is used to match the corresponding nodes in the existing medical education knowledge graph, and then the medical entities are Corresponding text snippet The relevant documents of the node are linked to the matched documents, and the content of the relevant documents of the node is updated; after all medical entities in each medical entity set are matched and linked, the linked medical education knowledge graph is obtained.

[0010] Furthermore, the step (3) specifically includes the following sub-steps: (3.1) Using a large language model, target any node in the medical education knowledge graph after the affiliation and its corresponding related literature Generate detailed description information; (3.2) Use a large language model to generate a node A concise summary of the text snippet on , the concise summary should include the main entity information and context; (3.3) Use the large language model to update the attributes of the edges in the affiliated medical education knowledge graph to obtain the updated medical education knowledge graph.

[0011] Furthermore, the step (4) is specifically as follows: applying the graph embedding algorithm to embed the updated medical education knowledge graph and generate a node vector set : ,in, represents any node vector, Represents a node vector set Chinese Communist Party node vectors, ; and the node vector collection Store in the atlas vector database.

[0012] Furthermore, the step (5) is specifically as follows: for each segmented text segment Apply text embedding method through vectorization model to embed representation and generate segmented text fragments Corresponding document vector set : ,in, represents any document vector, Represents a collection of document vectors Chinese Communist Party document vectors, ; Each generated document vector set is then stored in the text vector database.

[0013] Furthermore, the step (6) specifically includes the following sub-steps: (6.1) Perform entity recognition on the user's query and obtain the query entity set corresponding to the user's query : ,in, Represents any query entity obtained by entity recognition, Indicates that entity recognition is obtained query entities, ; (6.2) Each query entity Convert to query vector ; (6.3) Recall the query vector from the graph vector database using a similarity algorithm The relevant node vectors are Filtering and querying vectors Related node vectors, get the query vector The corresponding query node vector set ,in, is the similarity function, is the similarity threshold; (6.4) After recalling all query vectors, the total set of query node vectors corresponding to the user's query is obtained : .

[0014] Furthermore, the step (7) is specifically as follows: Convert the user's query Query into a user query vector ;according to Filter out the document set related to the user's query from the text fragment corresponding to each node vector in the total set of query node vectors ,in, It is the text fragment corresponding to any node vector in the total set of query node vectors.

[0015] Furthermore, the step (8) is specifically as follows: according to Filter out document sets related to the user's query in the text vector database .

[0016] The present invention also includes a knowledge graph-based retrieval enhancement generation device in a medical education scenario, including a memory and one or more processors, wherein the memory stores executable code, and when the one or more processors execute the executable code, it is used for the above-mentioned knowledge graph-based retrieval enhancement generation method in a medical education scenario.

[0017] The present invention also includes a computer-readable storage medium on which a program is stored. When the program is executed by a processor, it implements the above-mentioned knowledge graph retrieval enhancement generation method in a medical education scenario.

[0018] The beneficial effects of the present invention are as follows: By semantically segmenting medical textbooks and literature content and attaching them to a knowledge graph, the present invention achieves systematic management and dynamic updating of medical knowledge. Utilizing a large language model and graph embedding algorithm, the speed and accuracy of knowledge retrieval and reasoning are improved. Through entity recognition and similarity calculation methods, content relevant to user queries is filtered from the knowledge base, and accurate and logically rigorous answers are generated. This method significantly improves the level of teaching and scientific research in medical education, reduces manual workload, and has broad application prospects and practical value. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a flowchart of a knowledge graph-based retrieval enhancement generation method in a medical education scenario; Figure 2 This is a structural diagram of a knowledge graph-based retrieval enhancement generation device in a medical education scenario. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to illustrate the present invention, rather than to represent all embodiments. All other embodiments obtained by persons of ordinary skill in the art based on the embodiments of the present invention without creative work are within the scope of protection of the present invention.

[0021] Example 1: Figure 1 As shown, the present invention provides a knowledge graph-based retrieval enhancement generation method in a medical education scenario, comprising the following steps: (1) Use the document segmentation algorithm to semantically segment the text content of the medical textbook, ensuring that each segmented part is semantically independent and complete, and obtain N segmented text fragments: ,in, Represents any segmented text fragment, .

[0022] For example, a text content in a medical textbook is semantically segmented into multiple semantically independent text segments, each of which represents a complete medical concept or knowledge point.

[0023] (2) The N segmented text segments are attached to the existing medical education knowledge graph to obtain the attached medical education knowledge graph.

[0024] The step (2) specifically includes the following sub-steps: (2.1) Traverse each text segment after segmentation , perform named entity recognition and extract text fragments Medical entities in form medical entity sets : , Represents a text fragment Extractable medical entities, Represents a text fragment The extractable A medical entity.

[0025] For example, from the text segment “diabetes is a metabolic disease”, the medical entity “diabetes” is identified.

[0026] (2.2) Medical entity collection Each medical entity The entity linking technology is used to match the corresponding nodes in the existing medical education knowledge graph, and then the medical entities are Corresponding text snippet The relevant documents of the node are linked to the matched documents, and the content of the relevant documents of the node is updated; after all medical entities in each medical entity set are matched and linked, the linked medical education knowledge graph is obtained.

[0027] For example, the medical entity "diabetes" is matched with the corresponding node "diabetes" in the knowledge graph, and the relevant literature is attached to the node.

[0028] (3) Use the large language model technology to update the attributes of the nodes and edges in the affiliated medical education knowledge graph to obtain the updated medical education knowledge graph.

[0029] The step (3) specifically includes the following sub-steps: (3.1) Using a large language model, target any node in the medical education knowledge graph after the affiliation and its corresponding related literature Generate detailed description information , ensuring that the description covers the key attributes and background information of the entity.

[0030] For example, the relevant literature corresponding to the node "diabetes" generates detailed description information as "diabetes is a chronic metabolic disease, the main feature of which is high blood sugar level."

[0031] (3.2) Use a large language model to generate a node A concise summary of the text snippet on , so that key information can be quickly obtained when querying in the affiliated medical education knowledge graph, the concise summary should include the main entity information and context.

[0032] For example, the concise summary of the text segment attached to the node "diabetes" is "The main characteristic of diabetes is high blood sugar."

[0033] (3.3) Use the large language model to update the attributes of the edges in the affiliated medical education knowledge graph to obtain the updated medical education knowledge graph.

[0034] For example, the updated medical education knowledge graph is "Diabetes-Impact-Cardiovascular System".

[0035] (4) Apply the graph embedding algorithm to embed the updated medical education knowledge graph, generate a set of node vectors, and store them in the graph vector database.

[0036] The step (4) is specifically as follows: applying the graph embedding algorithm to embed the updated medical education knowledge graph and generate a node vector set : ,in, represents any node vector, Represents a node vector set Chinese Communist Party node vectors, ; and the node vector collection Store in the atlas vector database.

[0037] (5) Each of the N segmented text segments is embedded and represented through a vectorization model, and a document vector set corresponding to each segmented text segment is generated and stored in a text vector database.

[0038] The step (5) is specifically as follows: for each segmented text segment Apply text embedding method through vectorization model to embed representation and generate segmented text fragments Corresponding document vector set : ,in, represents any document vector, Represents a collection of document vectors Chinese Communist Party document vectors, ; Each generated document vector set is then stored in the text vector database.

[0039] (6) Perform entity recognition on the user's query and recall the relevant node vectors in the graph vector database to obtain the total set of query node vectors corresponding to the user's query .

[0040] Said step (6) specifically includes the following sub-steps: (6.1) Perform entity recognition on the user's query and obtain the query entity set corresponding to the user's query : ,in, Represents any query entity obtained by entity recognition, Indicates that entity recognition is obtained query entities, .

[0041] For example, entity recognition is performed on the user's query "complications of diabetes", and the query entity set obtained is "diabetes".

[0042] (6.2) Each query entity Convert to query vector .

[0043] (6.3) Recall the query vector from the graph vector database using a similarity algorithm The relevant node vectors are Filtering and querying vectors Related node vectors, get the query vector The corresponding query node vector set ,in, is the similarity function, is the similarity threshold.

[0044] (6.4) After recalling all query vectors, the total set of query node vectors corresponding to the user's query is obtained : .

[0045] (7) Using the text similarity calculation method, the document set related to the user's query is screened out from the text fragments corresponding to each node vector in the total set of query node vectors. .

[0046] The step (7) is specifically as follows: converting the user's query Query into a user query vector ;according to Filter out the document set related to the user's query from the text fragment corresponding to each node vector in the total set of query node vectors ,in, It is the text fragment corresponding to any node vector in the total set of query node vectors.

[0047] (8) Use the text similarity calculation method to directly filter out the document set related to the user's query in the text vector database .

[0048] The step (8) is specifically as follows: Filter out document sets related to the user's query in the text vector database .

[0049] (9) Based on the user's query, the total set of query node vectors , document collection and document collections , the answer corresponding to the query is generated through the large language model and returned to the user.

[0050] For example, the answer corresponding to the generated query Query is "Complications of diabetes include cardiovascular disease, kidney disease, retinopathy, etc."

[0051] This invention achieves high-quality open question-answering in medical education scenarios through systematic knowledge management and dynamic updating, combined with efficient retrieval and reasoning methods, effectively improving the level of teaching and scientific research.

[0052] Example 2: This example relates to a device for enhanced generation of retrieval based on knowledge graph in a medical education scenario, comprising a memory and one or more processors, wherein the memory stores executable code, and when the one or more processors execute the executable code, it is used for a method for enhanced generation of retrieval based on knowledge graph in a medical education scenario according to the above-mentioned Example 1; the device embodiment can be applied to any device with data processing capabilities, and the any device with data processing capabilities can be a device or apparatus such as a computer.

[0053] like Figure 2At the hardware level, the model watermark device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory. Of course, it may also include other hardware required for the business. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to achieve the above Figure 1 Of course, in addition to software implementation, the present invention does not exclude other implementation methods, such as logic devices or a combination of software and hardware, etc., that is, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0054] Improvements to a technology can be clearly categorized as either hardware improvements (for example, improvements to circuit structures like diodes, transistors, and switches) or software improvements (improvements to process flows). However, with technological advancements, many process flow improvements can now be considered direct improvements to hardware circuit structures. Designers almost always create the corresponding hardware circuit structure by programming the improved process flow into the hardware circuit. Therefore, it cannot be said that a process flow improvement cannot be implemented using physical hardware modules. For example, a programmable logic device (PLD), such as a field programmable gate array (FPGA), is an integrated circuit whose logical function is determined by user programming. Designers can integrate a digital system onto a PLD through their own programming, eliminating the need for a chip manufacturer to design and manufacture a dedicated integrated circuit chip. Moreover, nowadays, instead of manually fabricating integrated circuit chips, this programming is mostly performed using "logic compiler" software. This is similar to the software compiler used during program development. Before compilation, the original code must also be written in a specific programming language, called a hardware description language (HDL). There is not just one HDL, but many, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. The most commonly used ones are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art will also understand that by simply programming the method flow in one of these hardware description languages ​​and then programming it into an integrated circuit, a hardware circuit that implements the logic method flow can be easily obtained.

[0055] The controller can be implemented in any suitable manner. For example, the controller can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, an application-specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the memory control logic. Those skilled in the art will also appreciate that, in addition to implementing the controller purely in computer-readable program code, the controller can also be implemented in the form of logic gates, switches, an application-specific integrated circuit, a programmable logic controller, an embedded microcontroller, etc. by logically programming the method steps. Therefore, such a controller can be considered a hardware component, and the means for implementing the various functions included therein can also be considered as structures within the hardware component. Alternatively, the means for implementing the various functions can be considered both a software module implementing the method and a structure within the hardware component.

[0056] The systems, devices, modules, or units described in the above embodiments may be implemented by computer chips or entities, or by products having certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0057] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0058] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.

[0059] The present invention may be described in the general context of computer-executable instructions, such as program modules, executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, and the like that perform specific tasks or implement specific abstract data types. The present invention may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media, including storage devices.

[0060] Example 3: An embodiment of the present invention also provides a computer-readable storage medium on which a program is stored. When the program is executed by a processor, it implements the knowledge graph retrieval enhancement generation method based on a medical education scenario in the above-mentioned Example 1.

[0061] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A knowledge graph retrieval enhancement generation method in a medical education scenario, characterized in that: The following steps are involved: (1) Use the document segmentation algorithm to semantically segment the text content of the medical textbook, ensuring that each segmented part is semantically independent and complete, and obtain N segmented text fragments: ,in, represents any segmented text fragment, ; (2) attaching the N segmented text segments to the existing medical education knowledge graph to obtain the attached medical education knowledge graph; (3) Use the large language model technology to update the attributes of the nodes and edges in the medical education knowledge graph after the affiliation to obtain an updated medical education knowledge graph; (4) Apply the graph embedding algorithm to embed the updated medical education knowledge graph, generate a set of node vectors, and store them in the graph vector database; (5) The N segmented text segments are respectively embedded and represented by the vectorization model, and a document vector set corresponding to each segmented text segment is generated and stored in the text vector database; (6) Perform entity recognition on the user's query and recall relevant node vectors in the graph vector database to obtain the total set of query node vectors corresponding to the user's query; (7) Using a text similarity calculation method, a collection of documents related to the user's query is screened out from the text fragments corresponding to each node vector in the total set of query node vectors; (8) Using the text similarity calculation method, directly filter out the document set related to the user's query in the text vector database; (9) Based on the user's query, the total set of query node vectors, the document set and the document set, the large language model generates the answer corresponding to the query and returns it to the user.

2. According to claim 1, a knowledge graph retrieval enhancement generation method in a medical education scenario is characterized in that: The step (2) specifically includes the following sub-steps: (2.1) Traverse each segmented text segment , perform named entity recognition and extract text fragments Medical entities in form medical entity sets : ,in, Represents a text fragment Extractable medical entity, Represents a text fragment The extractable a medical entity; (2.2) Medical entity collection Each medical entity The medical entities are then matched with the corresponding nodes in the existing medical education knowledge graph through entity linking technology. The corresponding text fragment The relevant documents of the node are attached to the matched ones, and the content of the relevant documents of the node is updated; after all medical entities in each medical entity set are matched and attached, the attached medical education knowledge graph is obtained.

3. According to claim 2, a knowledge graph retrieval enhancement generation method in a medical education scenario is characterized in that: The step (3) specifically includes the following sub-steps: (3.1) Using a large language model, for any node in the medical education knowledge graph after the affiliation And its corresponding related literature Generate detailed description information; (3.2) Use a large language model to generate a node A concise summary of the text snippet on , the concise summary should include the main entity information and context; (3.3) Use the large language model to update the attributes of the edges in the affiliated medical education knowledge graph to obtain the updated medical education knowledge graph.

4. According to claim 3, a knowledge graph retrieval enhancement generation method in a medical education scenario is characterized in that: The step (4) is specifically as follows: applying a graph embedding algorithm to embed the updated medical education knowledge graph and generate a node vector set : ,in, represents any node vector, Represents a node vector collection Chinese Communist Party node vectors, ; and the node vector collection Stored in the atlas vector database.

5. According to claim 4, a knowledge graph retrieval enhancement generation method in a medical education scenario is characterized in that: The step (5) is specifically as follows: for each segmented text segment The text embedding method is applied through the vectorization model to embed the representation and generate the segmented text fragments The corresponding document vector set : ,in, represents any document vector, Represents a collection of document vectors Chinese Communist Party document vectors, ; Each generated document vector set is then stored in the text vector database.

6. According to claim 5, a knowledge graph retrieval enhancement generation method in a medical education scenario is characterized in that: The step (6) specifically includes the following sub-steps: (6.1) Perform entity recognition on the user's query Query and obtain the query entity set corresponding to the user's query Query : ,in, represents any query entity obtained by entity recognition, Indicates that entity recognition is obtained query entities, ; (6.2) For each query entity Convert to query vector ; (6.3) Recall the query vector from the graph vector database using a similarity algorithm The relevant node vectors are Filtering and querying vectors Related node vector, get the query vector The corresponding query node vector set ,in, is the similarity function, is the similarity threshold; (6.4) After recalling all query vectors, the total set of query node vectors corresponding to the user's query Query is obtained : .

7. According to claim 6, a knowledge graph retrieval enhancement generation method in a medical education scenario is characterized in that: The step (7) is specifically as follows: Convert the user's query Query into a user query vector ;according to Filter out the document set related to the user's query from the text fragment corresponding to each node vector in the total set of query node vectors ,in, It is the text fragment corresponding to any node vector in the total set of query node vectors.

8. According to claim 7, a knowledge graph-based retrieval enhancement generation method in a medical education scenario is characterized in that: The step (8) is specifically as follows: according to Filter out the document collection related to the user's query in the text vector database .

9. A knowledge graph-based retrieval enhancement generation device in a medical education scenario, characterized in that: It includes a memory and one or more processors, wherein the memory stores executable code, and when the one or more processors execute the executable code, it is used to implement the knowledge graph retrieval enhancement generation method based on a medical education scenario as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that: A program is stored thereon, and when the program is executed by a processor, it implements a knowledge graph retrieval enhancement generation method based on a medical education scenario as described in any one of claims 1-8.

Images