Method and system for constructing medical knowledge graph based on medical ancient books

Abstract

The invention relates to a method and system for constructing a medical knowledge graph based on medical ancient books, and belongs to the technical field of medical ancient book processing, and the method comprises the following steps: carrying out the entity recognition and labeling of a traditional Chinese medicine ancient book text, and obtaining an entity set with type labeling; performing relation extraction on the entity set to obtain an entity relation triple set; constructing a knowledge graph based on the entity set, the entity relationship triple set and a standard ontology; performing semantic indexing and association on the knowledge graph to obtain a knowledge graph after semantic enhancement; and performing evaluation, verification and optimization on the knowledge graph after semantic enhancement to obtain a final optimized knowledge graph. According to the method, on the basis of understanding the new medical literature, the characteristics of the new medical literature are combined, the natural language processing technology is utilized, and the characteristics of themes, contents, subject fields and the like are analyzed and perfected through text word segmentation and entity recognition, so that the corresponding classification keyword table is generated.

Description

Technical Field

[0001] This disclosure belongs to the field of medical ancient book processing technology, and in particular relates to a method and system for constructing a medical knowledge graph based on medical ancient books. Background Technology

[0002] The medical literature of Xin'an is renowned worldwide for its vast quantity, high academic value, and significant social impact. The *Xin'an Medical Classics Examination* records 835 works, covering 13 categories of medical texts. Its comprehensiveness and richness of content make it a standout among regionally named schools of thought. Currently, approximately 500 titles survive, becoming a valuable resource for the study of Xin'an medicine and a unique resource and advantage of traditional Chinese medicine.

[0003] Currently, research is focused on regional schools of traditional Chinese medicine, establishing a digital platform for related literature, and conducting research on literature mining and utilization, especially for key diseases—a precedent in China. Based on the deep integration of Xin'an medicine with modern technology, this research aims to create a digital knowledge system for Xin'an medicine, establish an open, shared, systematic, and multi-dimensional literature data platform, and realize "Digital Xin'an Medicine." This will provide digital resource-level knowledge services for the research, inheritance, and development of Xin'an medicine, transforming historical documents into real productivity and fully realizing the multiple values ​​of literature resources. This not only fills a gap in Xin'an medicine research but also aligns with the research direction and tasks of the Institute of Xin'an Medicine and Modernization of Traditional Chinese Medicine, possessing significant scientific and technological value.

[0004] Therefore, it is necessary to provide a new method and system for constructing medical knowledge graphs based on ancient medical texts to solve the above-mentioned technical problems. Summary of the Invention

[0005] The purpose of this disclosure is to provide a method and system for constructing medical knowledge graphs based on ancient medical texts in order to solve the above-mentioned problems.

[0006] This disclosure achieves the above objectives through the following technical solutions: A method for constructing a medical knowledge graph based on ancient medical texts includes the following steps: Entity recognition and annotation are performed on ancient Chinese medicine texts to obtain a set of entities with type annotations; Relation extraction is performed on the entity set to obtain a set of entity-relation triples; A knowledge graph is constructed based on the entity set, the entity relation triplet set, and the standard ontology; The knowledge graph is semantically indexed and associated to obtain a semantically enhanced knowledge graph. The semantically enhanced knowledge graph is evaluated, verified, and optimized to obtain the optimized final knowledge graph.

[0007] As a further optimization of this disclosure, entity recognition and annotation are performed on the original ancient text to obtain a set of entities with type annotations, including: Data preprocessing of ancient Chinese medicine texts; A dictionary of traditional Chinese medicine terminology was constructed based on preprocessed ancient Chinese medicine texts. The rule-based annotation method performs preliminary entity annotation on the preprocessed ancient Chinese medicine texts based on the Chinese medicine terminology dictionary to obtain preliminary annotation results; The preliminary annotation results are optimized using machine learning annotation to obtain structured entity annotation results, i.e., a set of entities with type annotations.

[0008] As a further optimization of this disclosure, relation extraction is performed on the entity set to obtain a set of entity relation triples, including: Relationship types are defined based on preset rules; Based on the deep learning model, the type-labeled entity set, and the relation type, relation classification, entity recognition, and relation extraction are performed on sentences or context fragments containing target entities to obtain a structured set of entity relation triples.

[0009] As a further optimization of this disclosure, a knowledge graph is constructed based on the entity set, the entity relation triplet set, and the standard ontology, including: Map the entity set to a standard ontology; Based on the defined relation types and the entity set, design the node types, relation types and their attributes of the knowledge graph, and construct the knowledge graph.

[0010] As a further optimization of this disclosure, the knowledge graph is semantically indexed and associated to obtain a semantically enhanced knowledge graph, including: The entities in the knowledge graph are compared with standard ontologies to establish mapping relationships; Identify identical or similar entities in different ancient Chinese medicine books and link them together in a knowledge graph to form a unified node; Create cross-document entity association edges in the knowledge graph and record the source and similarity information of the associated documents; By combining entity standard mapping and cross-document association results, semantic indexing is performed on knowledge graph nodes and edges, and semantic attributes of entities are added to obtain a semantically enhanced knowledge graph.

[0011] As a further optimization of this disclosure, the semantically enhanced knowledge graph is evaluated, verified, and optimized to obtain an optimized final knowledge graph, including: The semantically enhanced knowledge graph is subjected to data quality assessment and verification; Based on the evaluation metrics and verification results, the entity recognition rules and relation extraction model parameters are adjusted, and the knowledge graph construction and semantic indexing strategies are optimized to obtain the optimized final knowledge graph.

[0012] A system for constructing a medical knowledge graph based on ancient medical texts includes: The entity recognition and annotation module is used to perform entity recognition and annotation on ancient Chinese medicine texts, and obtain a set of entities with type annotations; The relation extraction module is used to extract relations from the entity set to obtain a set of entity relation triples. The knowledge graph construction module is used to construct a knowledge graph based on the entity set, the entity relation triplet set, and the standard ontology. The semantic indexing and association module is used to perform semantic indexing and association on the knowledge graph to obtain a semantically enhanced knowledge graph; The optimization module is used to evaluate, verify, and optimize the semantically enhanced knowledge graph to obtain the optimized final knowledge graph.

[0013] As a further optimization of this disclosure, the entity recognition and annotation module performs entity recognition and annotation on the original ancient text to obtain a set of entities with type annotations, including: Data preprocessing of ancient Chinese medicine texts; A dictionary of traditional Chinese medicine terminology was constructed based on preprocessed ancient Chinese medicine texts. The rule-based annotation method performs preliminary entity annotation on the preprocessed ancient Chinese medicine texts based on the Chinese medicine terminology dictionary to obtain preliminary annotation results; The preliminary annotation results are optimized using machine learning annotation to obtain structured entity annotation results, i.e., a set of entities with type annotations.

[0014] An electronic device includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; The processor is used to execute the program stored in the memory to implement the method of constructing a medical knowledge graph based on ancient medical books.

[0015] A computer-readable storage medium storing a computer program, characterized in that, when the computer program is executed by a processor, it implements the method for constructing a medical knowledge graph based on ancient medical texts.

[0016] The beneficial effects of this disclosure are as follows: Based on an understanding of Xin'an medical literature and its unique characteristics, this publication utilizes natural language processing technology, including text segmentation and entity recognition, to analyze and refine the literature's themes, content, and subject areas, generating a corresponding knowledge graph (classified keyword table). This knowledge graph (classified keyword table) completes the formulation of search keywords, keyword synonyms, and hypernyms for the Xin'an medical database platform. The construction of the knowledge graph (classified keyword table) serves the database's construction needs. Attached Figure Description

[0017] Figure 1 This is a flowchart of a method in an embodiment of this disclosure; Figure 2 This is a system structure block diagram of an embodiment of this disclosure; Figure 3 This is a block diagram of the device structure in an embodiment of this disclosure. Detailed Implementation

[0018] The present application will now be described in further detail with reference to the accompanying drawings. It should be noted that the following specific embodiments are only used to further illustrate the present application and should not be construed as limiting the scope of protection of the present application. Those skilled in the art can make some non-essential improvements and adjustments to the present application based on the above application content.

[0019] like Figure 1 As shown, a method for constructing a medical knowledge graph based on ancient medical texts includes the following steps: S1. Perform entity recognition and annotation on ancient Chinese medicine texts to obtain a set of entities with type annotations; specifically including: S11: Input the corpus of ancient Chinese medicine texts.

[0020] S12: Data preprocessing. Clean the text data, remove irrelevant characters and punctuation marks, and perform sentence segmentation.

[0021] S13: Construct a dictionary of TCM terminology. This includes a list of entity words for medicinal materials, symptoms, and prescriptions.

[0022] S14: Rule-based annotation method, which performs preliminary entity annotation on the text based on a terminology dictionary, and obtains preliminary annotation results, including: S141: Construct a dictionary of terms in the field of traditional Chinese medicine, such as Chinese herbal medicine, prescriptions, diagnostic methods, and theories of traditional Chinese medicine.

[0023] S142: Using string matching methods, perform preliminary entity annotation on the text based on a terminology dictionary to obtain preliminary annotation results.

[0024] S143: Scan the text, mark strings that match dictionary entries and rules, and assign them predefined type labels, such as ["etiology", "pathogenesis"] - traditional Chinese medicine theory; ["ephedra", "cinnamon twig"] - traditional Chinese medicine.

[0025] S15: Machine Learning Labeling Optimization S151: Use sequence labeling models (such as BiLSTM-CRF, BERT-CRF, BERT-BiLSTM-CRF).

[0026] S152: Train the model on a high-quality annotated corpus of ancient Chinese medicine texts to learn entity boundaries and type features.

[0027] S153: Model Input: The preliminary annotation results are used as auxiliary input, i.e., the sentence sequence after character / word segmentation. Combined with the model prediction results, the entity recognition accuracy is optimized.

[0028] S154: Model output: The label sequence corresponding to each character / word (such as BIO, BIOES format), indicating the entity boundary (B-Begin, I-Inside, E-End, S-Single) and entity type (such as B-traditional Chinese medicine, I-traditional Chinese medicine, B-formula).

[0029] S16: Output structured entity annotation results. Each entity contains: original text fragment, start position, end position, entity type (e.g., Chinese medicine, prescription, acupuncture and massage, diagnosis method, treatment method. For example: [Text: "Unclear cough is due to lung dysfunction", entity: {"cough": [position: 0-3, type: internal medicine], "etiology and pathogenesis": [position: 6-8, type: diagnosis method]}).

[0030] S2. Extract relations from the entity set to obtain a set of entity relation triples; specifically including: Based on the identified entities and their contextual information, specific semantic relationships between the entities are extracted. Relationship definitions: "Internal medicine - includes - disease syndrome" (e.g., "pulmonary diseases syndrome - includes - cough"), "formula - has - efficacy" (e.g., "Yigongsan - has - tonic").

[0031] S21: Input the entity annotation results of S1.

[0032] S22: Define relation types based on rules. Based on common sense and domain knowledge, determine the relation types to be extracted, such as "prescription-contains-medicinal materials" and "symptom-applicable-prescription", including: S221: Analyze common sentence patterns expressing specific relationships in the corpus. For example, for "prescription-containing-traditional Chinese medicine": "[prescription] is composed of [medicinal material 1], [medicinal material 2], ..."; for "symptom-applicable-prescription": "[symptom] patients should use [prescription]" or "[prescription] is the main treatment".

[0033] S222: Using sentence templates (such as "xx Fang treats xx") and dependency parsing, the entities identified by S1 fill the slots in the template to form triples.

[0034] S23: Deep learning-based methods: S231: Using the (entity + relation) corpus annotated by S1, locate entity pairs in sentences (usually adjacent entities or entities in a specific grammatical relation).

[0035] S232: Perform relation classification on sentences or context fragments containing target entity pairs. Specifically, use a text classification model (such as BERT, TextCNN, TextRNN), with the input being sentences containing entity tags (such as those marked with special symbols), and the output being a predefined relation type or "no relation".

[0036] S233: The model performs entity recognition and relation extraction simultaneously. Examples include CasRel (based on BERT, first extracting subject entities, then extracting object entities and relations for each subject entity), TPLinker (modeling relations as token links between entity pairs), and SPN (generating triples from sequence to sequence).

[0037] S24: Output a structured set of relation triples. Each triple contains: "head entity", "relation type", and "tail entity". For example: <Guizhi Tang, contains, Guizhi>, <Shanghan, applicable, Mahuang Tang>.

[0038] S3. Construct a knowledge graph based on the entity set, the entity relation triplet set, and the standard ontology; specifically including: The extracted entities and relationships are organized into a structured knowledge graph and mapped to domain-standard terminology to achieve semantic consistency and cross-document association.

[0039] S31: Ontology Mapping (Semantic Standardization). This involves mapping the entities identified in S1 (especially core concepts such as diagnostic methods, prescriptions, and traditional Chinese medicine) to authoritative ontology or standard terminology systems (such as GB / T15657-2021 "Classification and Codes of Diseases in Traditional Chinese Medicine", standard names in the textbook "Traditional Chinese Materia Medica", and the list of classic prescriptions in "Formulary"). This includes: S311: Based on string similarity (edit distance, Jaccard).

[0040] S312: Semantic similarity calculation based on word vector / ontology embedding.

[0041] S313: Rule matching (alias and synonym list) solves the problems of polysemy and homosemantic relationships, unifies knowledge representation, and lays the foundation for subsequent association and reasoning. For example, it maps "blood residue charcoal" in ancient books to the standard term "human hair ash"; and maps "inner court" to "leading court".

[0042] S32: Knowledge Graph Schema Definition: S321: Based on the defined relation types (S2) and entity types (S1), design the node types (entity types) and relation types (edge ​​types) of the graph, along with their attributes. For example: node type, "Diagnosis Method" (attributes: etiology, pathogenesis...). S322: Relationship type, "Contains" (Traditional Chinese Medicine -> Heat-clearing Medicine), "Applicable" (Heat-clearing Medicine -> Isatis Root), "Has" (Isatis Root -> Heat-clearing), "Belongs to" (Internal Medicine -> Disease / Syndrome).

[0043] S33: Map Construction and Storage: S331: Use entities that have been standardized by ontology mapping as nodes in the knowledge graph.

[0044] S332: Use the "relation triples" extracted from S2 as "edges" of the knowledge graph.

[0045] S333: Additional information about entities and relations (such as the original text fragments, locations, and confidence levels identified by S1; the source sentences of relations extracted by S2; the standard IDs of ontology mapping, etc.) are used as "attributes" of nodes or edges.

[0046] S34: Output the initially constructed knowledge graph file and store it in Neo4j graph database format.

[0047] S4. Semantically index and associate the knowledge graph to obtain a semantically enhanced knowledge graph; specifically including: S41: Input: S34 preliminary knowledge graph file.

[0048] S42: Entity Normalization Mapping S421: Compare entities in the knowledge graph with standard ontologies (such as GB / T15657-2021 "Classification and Codes of Diseases in Traditional Chinese Medicine") to establish mapping relationships. For example, map "stroke" in ancient books to the standard term "apoplexy".

[0049] S422: This unified node links to its location in different documents, contextual descriptions, related prescriptions / medicinal materials / symptoms, and other information. Update the attribute information of the knowledge graph nodes, adding standard terminology identifiers and mapping relationships.

[0050] S43: Cross-document entity association: S431: Identify the same or similar entities in different ancient books, and link the "same standard entity" (such as "Guizhi Tang" and "Taiyang Bing") mentioned in different ancient books (such as "Chishui Xuanzhu", "Bencao Beiyao", and "Mingyi Lei'an") in the knowledge graph to form a unified node.

[0051] S432: Create cross-document entity association edges in the knowledge graph and record the source and similarity information of the associated documents.

[0052] S44: Enhanced Semantic Indexing S441: Integrate entity standard mapping and cross-document association results to semantically index knowledge graph nodes and edges, enriching the semantic information of entities and relationships.

[0053] S442: Add semantic attributes such as entity aliases, classification levels, and functional indications, and improve the semantic descriptions of relationships, including applicable scenarios and basis. S45: Output: A semantically enhanced knowledge graph file containing standard entity mappings, cross-document associations, and detailed semantic indexing information.

[0054] S5. Evaluate, verify, and optimize the semantically enhanced knowledge graph to obtain the optimized final knowledge graph; specifically including: Ensure the accuracy, consistency, coverage, and quality of the final map results at each stage of processing.

[0055] S51: Input the semantically enhanced knowledge graph file; S52: Data Quality Assessment S521: Calculate precision, recall, and F1 score to evaluate the accuracy of entity recognition and relation extraction.

[0056] S522: Statistical term coverage (the proportion of indexed entities to actual entities), measures the comprehensiveness of entity identification.

[0057] S523: Analyze the semantic consistency and logical coherence of entities and relations in a knowledge graph.

[0058] S53: Development of a Manual Verification Platform S531: Construct a Web annotation system to display entity, relation, and semantic indexing information in the knowledge graph for experts to correct and supplement.

[0059] S532: Implement a double-blind verification process, where different experts independently verify the same content and compare the results to ensure consistency.

[0060] S54: Expert Verification and Feedback S541: Organize experts in the field of traditional Chinese medicine to use the verification platform to conduct a comprehensive review of the knowledge graph.

[0061] S542: Collect expert feedback, correct erroneous entities, relationships, and semantic indexes, and supplement missing information.

[0062] S55: Quality Iterative Optimization S551: Based on the evaluation indicators and expert verification results, adjust the entity recognition rules, relation extraction model parameters, and optimize the knowledge graph construction and semantic indexing strategies.

[0063] S552: Repeat the above steps to continuously improve data quality until the expected standard is reached.

[0064] S56: Output the optimized final knowledge graph.

[0065] In this embodiment, the knowledge graph includes a classification vocabulary, which is generated as follows: 1. Starting from the theoretical knowledge and application of Xin'an medicine, this study uses 15 literature knowledge categories as primary category headings, including Traditional Chinese Medicine theory, Typhoid Fever and Golden Chamber Prescriptions, febrile diseases, diagnostic methods, Chinese herbal medicine, prescriptions, internal medicine, surgery, gynecology, pediatrics, ENT, bone and joint injuries, acupuncture and massage, health preservation, and others. Within each knowledge category, the superordinate terms of keywords are used as secondary category headings. Guided by these secondary headings, keywords are formed within the content system of the secondary headings. Further expansion is then made to create multiple synonyms for the keywords. These synonyms are semantically similar to the keywords, ensuring comprehensive terminology coverage when using keywords, superordinate terms, and synonyms as search terms. This systematically sorts out and accurately models the terminology system of Xin'an medicine literature, satisfying comprehensive literature searches under different search terms and reducing the possibility of literature omissions.

[0066] Xin'an Medical Knowledge Indexing Classification Thesaurus 1: Keyword List of Xin'an Medical Disciplines and Theories

[0067] 2. Based on the advantageous diseases of Xin'an Medicine, and building upon the Xin'an Medicine Thesaurus, five diseases—lung diseases, brain diseases, arthralgia, diabetes, and epidemic diseases—are used as primary category headings. Specific disease names under each disease category are used as secondary category headings. Under the guidance of these secondary headings, keywords are formed within the content of each secondary heading. Through disease classification and semantic expansion, multiple keywords are generated, further enriching the thesaurus's universality and scientific rigor.

[0068] Xin'an Medical Knowledge Indexing Classification Thesaurus 2: Keyword List for Five Major Advantageous Diseases

[0069] Using this thesaurus as the indexing rule, the machine indexing of Xin'an medical literature is completed, ensuring the coverage of literature content during the user's search process.

[0070] This thesaurus, based on the *Thesaurus of Traditional Chinese Medicine* published by the Institute of Library and Information Science, China Academy of Traditional Chinese Medicine, was refined through careful reading and understanding of Xin'an medical literature. It incorporates the unique characteristics of Xin'an medical literature and utilizes Natural Language Processing (NLP) technology, including text segmentation and entity recognition, to analyze its themes, content, and disciplinary fields, generating a corresponding categorized keyword list. This categorized thesaurus completed the formulation of search keywords, keyword synonyms, and hypernyms for the Xin'an medical database platform, yielding 4195 keywords, 7603 keyword synonyms, and 6972 keyword hypernyms. The construction of the categorized thesaurus serves the needs of database construction; the initial goal is to build "one master database and one sub-database," namely, a master database for Xin'an medical data and a sub-database for specific diseases.

[0071] like Figure 2 As shown, embodiments of this disclosure provide a system for constructing a medical knowledge graph based on ancient medical texts, including: The entity recognition and annotation module is used to perform entity recognition and annotation on ancient Chinese medicine texts, and obtain a set of entities with type annotations; The relation extraction module is used to extract relations from the entity set to obtain a set of entity relation triples. The knowledge graph construction module is used to construct a knowledge graph based on the entity set, the entity relation triplet set, and the standard ontology. The semantic indexing and association module is used to perform semantic indexing and association on the knowledge graph to obtain a semantically enhanced knowledge graph; The optimization module is used to evaluate, verify, and optimize the semantically enhanced knowledge graph to obtain the optimized final knowledge graph.

[0072] The implementation process of the functions and roles of each module in the above system is detailed in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0073] For the system embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The system embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0074] In the above embodiments, any number of modules can be combined into one module, or any one module can be split into multiple modules. Alternatively, at least some of the functionality of one or more of these modules can be combined with at least some of the functionality of other modules and implemented in one module. At least one of all modules can be at least partially implemented as hardware circuitry, such as a Field Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), a System-on-Chip, a System-on-Substrate, a System-on-Package, an Application-Specific Integrated Circuit (ASIC), or any other reasonable means of integrating or packaging circuitry, or implemented in software, hardware, or firmware, or in any suitable combination of any of these three implementation methods. Alternatively, at least one of all modules can be at least partially implemented as a computer program module, which, when run, can perform corresponding functions.

[0075] See Figure 3 The electronic device provided in the embodiments of this disclosure includes a processor 1110, a communication interface 1120, a memory 1130 and a communication bus 1140, wherein the processor 1110, the communication interface 1120 and the memory 1130 communicate with each other through the communication bus 1140. Memory 1130 is used to store computer programs; The processor 1110, when executing the program stored in the memory 1130, implements the above-described method for constructing a medical knowledge graph based on ancient medical texts. The aforementioned communication bus 1140 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus 1140 can be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, it is represented by only one thick line in the figure, but this does not indicate that there is only one bus or one type of bus.

[0076] The communication interface 1120 is used for communication between the above-mentioned electronic device and other devices.

[0077] The memory 1130 may include random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Optionally, the memory 1130 may also be at least one storage device located remotely from the aforementioned processor 1110.

[0078] The processor 1110 mentioned above can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0079] Embodiments of this disclosure also provide a computer-readable storage medium. The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method described above for constructing a medical knowledge graph based on ancient medical texts.

[0080] The computer-readable storage medium may be included in the device / apparatus described in the above embodiments; or it may exist independently and not assembled into the device / apparatus. The computer-readable storage medium carries one or more programs that, when executed, implement the method for constructing a medical knowledge graph based on ancient medical texts according to embodiments of this disclosure.

[0081] According to embodiments of this disclosure, the computer-readable storage medium can be a non-volatile computer-readable storage medium, such as including, but not limited to: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this disclosure, the computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0082] The embodiments described above are merely examples of several implementations of this disclosure, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent disclosure. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this disclosure, and these modifications and improvements all fall within the protection scope of this disclosure.

Claims

1. A method for constructing a medical knowledge graph based on ancient medical texts, characterized in that, Includes the following steps: Entity recognition and annotation are performed on ancient Chinese medicine texts to obtain a set of entities with type annotations; Relation extraction is performed on the entity set to obtain a set of entity-relation triples; A knowledge graph is constructed based on the entity set, the entity relation triplet set, and the standard ontology; The knowledge graph is semantically indexed and associated to obtain a semantically enhanced knowledge graph. The semantically enhanced knowledge graph is evaluated, verified, and optimized to obtain the optimized final knowledge graph.

2. The method for constructing a medical knowledge graph based on ancient medical texts according to claim 1, characterized in that, Entity recognition and annotation are performed on the original ancient texts to obtain a set of entities with type annotations, including: Data preprocessing of ancient Chinese medicine texts; A dictionary of traditional Chinese medicine terminology was constructed based on preprocessed ancient Chinese medicine texts. The rule-based annotation method performs preliminary entity annotation on the preprocessed ancient Chinese medicine texts based on the Chinese medicine terminology dictionary to obtain preliminary annotation results; The preliminary annotation results are optimized using machine learning annotation to obtain structured entity annotation results, i.e., a set of entities with type annotations.

3. The method for constructing a medical knowledge graph based on ancient medical texts according to claim 1, characterized in that, Relation extraction is performed on the entity set to obtain a set of entity-relation triples, including: Relationship types are defined based on preset rules; Based on the deep learning model, the type-labeled entity set, and the relation type, relation classification, entity recognition, and relation extraction are performed on sentences or context fragments containing target entities to obtain a structured set of entity relation triples.

4. The method for constructing a medical knowledge graph based on ancient medical texts according to claim 1, characterized in that, A knowledge graph is constructed based on the entity set, the entity relation triplet set, and the standard ontology, including: Map the entity set to a standard ontology; Based on the defined relation types and the entity set, design the node types, relation types and their attributes of the knowledge graph, and construct the knowledge graph.

5. The method for constructing a medical knowledge graph based on ancient medical texts according to claim 1, characterized in that, The knowledge graph is semantically indexed and associated to obtain a semantically enhanced knowledge graph, including: The entities in the knowledge graph are compared with standard ontologies to establish mapping relationships; Identify identical or similar entities in different ancient Chinese medicine books and link them together in a knowledge graph to form a unified node; Create cross-document entity association edges in the knowledge graph and record the source and similarity information of the associated documents; By combining entity standard mapping and cross-document association results, semantic indexing is performed on knowledge graph nodes and edges, and semantic attributes of entities are added to obtain a semantically enhanced knowledge graph.

6. The method for constructing a medical knowledge graph based on ancient medical texts according to claim 1, characterized in that, The semantically enhanced knowledge graph is evaluated, verified, and optimized to obtain the optimized final knowledge graph, including: The semantically enhanced knowledge graph is subjected to data quality assessment and verification; Based on the evaluation metrics and verification results, the entity recognition rules and relation extraction model parameters are adjusted, and the knowledge graph construction and semantic indexing strategies are optimized to obtain the optimized final knowledge graph.

7. A system for constructing a medical knowledge graph based on ancient medical texts, characterized in that, include: The entity recognition and annotation module is used to perform entity recognition and annotation on ancient Chinese medicine texts, and obtain a set of entities with type annotations; The relation extraction module is used to extract relations from the entity set to obtain a set of entity relation triples. The knowledge graph construction module is used to construct a knowledge graph based on the entity set, the entity relation triplet set, and the standard ontology. The semantic indexing and association module is used to perform semantic indexing and association on the knowledge graph to obtain a semantically enhanced knowledge graph; The optimization module is used to evaluate, verify, and optimize the semantically enhanced knowledge graph to obtain the optimized final knowledge graph.

8. A system for constructing a medical knowledge graph based on ancient medical texts according to claim 7, characterized in that, The entity recognition and annotation module performs entity recognition and annotation on the original ancient text, obtaining a set of entities with type annotations, including: Data preprocessing of ancient Chinese medicine texts; A dictionary of traditional Chinese medicine terminology was constructed based on preprocessed ancient Chinese medicine texts. The rule-based annotation method performs preliminary entity annotation on the preprocessed ancient Chinese medicine texts based on the Chinese medicine terminology dictionary to obtain preliminary annotation results; The preliminary annotation results are optimized using machine learning annotation to obtain structured entity annotation results, i.e., a set of entities with type annotations.

9. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor for executing a program stored in a memory to implement the method for constructing a medical knowledge graph based on ancient medical texts as described in any one of claims 1-6.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for constructing a medical knowledge graph based on ancient medical texts as described in any one of claims 1-6.

Images