A knowledge service platform for inheritance and popularization of folk bone-setting methods

Abstract

The application relates to the technical field of electric digital data processing, in particular to a knowledge service platform for inheritance and popularization of folk bone-setting prescriptions, corresponding steps comprising: constructing core nodes and basic topological structures of a semantic field based on a standard knowledge system; constructing a mapping relationship between text and nodes and an extended node to obtain a semantic association path in the basic topological structure according to the similarity between included semantic fragments and the core nodes; constructing a relationship edge in the semantic association path based on the co-occurrence relationship between the included semantic fragments, determining the weight value of the relationship edge based on the source of the relationship edge, and obtaining a target semantic field; screening candidate nodes to obtain an archived knowledge subgraph based on the candidate node set of the unrecorded semantic fragments in the target semantic field, the relationship edges between the candidate nodes and the weight values of the relationship edges, so as to archive folk bone-setting prescription knowledge; and the technical scheme provides reliable technical support for systematic management, accurate inheritance and popularization and value reservation of the folk bone-setting prescriptions.

Description

Technical Field

[0001] This invention relates to the field of electronic digital data processing technology, specifically to a knowledge service platform for the inheritance and promotion of traditional bone-setting remedies. Background Technology

[0002] Folk bone-setting remedies, as an important component of traditional Chinese medicine orthopedics, are widely circulated throughout my country. These "traditional prescriptions" originating from clinical practice have unique efficacy in fracture healing. Currently, research on folk bone-setting remedies mainly focuses on case collection and small-scale clinical observation, lacking a systematic knowledge management system. This is due to the differences in the composition, preparation process, and dosage of folk bone-setting remedies across different regions. Furthermore, these remedies often exist in the form of oral accounts, handwritten copies, or local documents, lacking unified knowledge expression and input standards, resulting in inconsistent data quality. This unstructured and non-standardized knowledge form fragments the overall knowledge. Existing technologies generally break down folk bone-setting remedies into discrete fields, failing to retain the core logic of folk bone-setting knowledge and easily causing ambiguity and confusion. Establishing a digital knowledge base would diminish the original value of folk knowledge. Summary of the Invention

[0003] To address the technical problems of significant knowledge fragmentation and value loss when digitizing existing folk bone-setting knowledge, the present invention aims to provide a knowledge service platform for the inheritance and promotion of folk bone-setting techniques. The specific technical solution adopted is as follows: This invention provides a knowledge service platform for the inheritance and promotion of traditional folk bone-setting remedies, the platform comprising: The semantic field construction module is used to construct the core nodes of the semantic field and the basic topology between the core nodes based on the standard knowledge system. According to the similarity between the semantic fragments in the collected folk bone-setting prescription texts and the core nodes, the module constructs the mapping relationship between text and nodes in the basic topology and expands the nodes to obtain semantic association paths. Based on the co-occurrence relationship between the collected semantic fragments in the collected folk bone-setting prescription texts, the module constructs relation edges in the semantic association paths and determines their weight values ​​based on the source of the relation edges to obtain the target semantic field. The knowledge archiving module is used to filter candidate nodes in the target semantic field based on the set of candidate nodes in the unrecorded semantic fragments in the unrecorded folk bone-setting prescription texts, the relationship edges between the candidate nodes and the weight values ​​of the relationship edges, to obtain an archived knowledge subgraph for archiving folk bone-setting prescription knowledge.

[0004] Furthermore, the core nodes of the semantic field constructed based on the standard knowledge system and the basic topological structure between the core nodes include: Based on the standard knowledge system of traditional Chinese medicine, each standard knowledge entity in it is used as the corresponding core node in the semantic field. Based on the knowledge types of the core nodes, basic relationship types between the core nodes are constructed, and a basic topology is formed based on the core nodes and basic relationship types.

[0005] Furthermore, based on the similarity between semantic fragments and core nodes in the collected folk bone-setting prescriptions, a mapping relationship between text and nodes is constructed in the basic topology, including: Using a pre-defined deep learning model, we can identify the semantic fragments included in the existing folk bone-setting prescriptions texts and determine the target core nodes in the basic topology that correspond to the target semantic fragments. Determine the cosine similarity between the corresponding vectors of the target semantic segments and the target core nodes, and establish a mapping relationship between the target semantic segments and the target core nodes whose cosine similarity is greater than or equal to a preset similarity threshold.

[0006] Furthermore, the step of constructing a text-node mapping relationship and expanding nodes to obtain semantic association paths in the basic topology structure based on the similarity between semantic fragments and core nodes in the collected folk bone-setting prescription texts includes: Semantic fragments with a cosine similarity less than a preset similarity threshold are created as extended nodes in the basic topology. Based on the semantic approximation relationship between the extended nodes and the core nodes, the connection relationship between the extended nodes and the core nodes is constructed, and the semantic association path is obtained based on the mapping relationship and the connection relationship.

[0007] Furthermore, based on the co-occurrence relationships between the included semantic fragments in the already included folk bone-setting prescription texts, relational edges are constructed in the semantic association path, including: Determine the point-like mutual information representing the co-occurrence relationship between any two semantic fragments in the collected collection of folk bone-setting prescriptions; For two included semantic fragments whose point-based mutual information is greater than a preset relevance threshold, a relationship edge is constructed between them in the semantic association path.

[0008] Furthermore, the weight values ​​of relation edges are determined based on their origin, including: The initial weight value of the relation edge is determined based on the source authority evaluation, the weight increase is determined based on the verification frequency of the relation edge, and the updated weight value of the relation edge is obtained by updating the initial weight value using the weight increase.

[0009] Furthermore, based on the candidate node set in the target semantic field of unrecorded semantic fragments from unrecorded folk bone-setting prescription texts, the relationship edges between candidate nodes, and the weight values ​​of these relationship edges, candidate nodes are filtered to obtain an archived knowledge subgraph, including: Identify the target unrecorded semantic fragments in the texts of unrecorded folk bone-setting remedies, and retrieve the candidate nodes corresponding to the target unrecorded semantic fragments in the target semantic field to obtain a set of candidate nodes; Based on the candidate node set and the relationship edges between the candidate nodes and their weights, a fuzzy knowledge subgraph of uncollected folk bone-setting prescriptions is obtained. Candidate nodes are then filtered to obtain the corresponding archived knowledge subgraph.

[0010] Furthermore, the process of filtering candidate nodes to obtain the corresponding archived knowledge subgraph includes: Candidate nodes are filtered using preset constraint rules to obtain the archived knowledge subgraph corresponding to the fuzzy knowledge subgraph. The preset constraint rules include at least one or more of the following: The first constraint rule is that there is a direct relationship edge between adjacent candidate nodes in the target semantic field; the second constraint rule is that the node types corresponding to the candidate nodes have a necessary relational dependency. The first and second constraint rules are used to filter out candidate nodes in the candidate node set that do not meet the rules.

[0011] Furthermore, the process of filtering candidate nodes to obtain the corresponding archived knowledge subgraph includes: Determine the target relation path formed by relation edges between any candidate nodes corresponding to any two unrecorded semantic segments; By using the weights and values ​​of all relation edges on the target relation path, the path strength of the target relation path is determined, and the connection strength between the corresponding candidate nodes is obtained. Based on the connection strength, the connectivity of the target candidate node is determined, and the candidate node with the lowest connectivity is iteratively eliminated to obtain the filtered candidate nodes and the archived knowledge subgraph corresponding to the fuzzy knowledge subgraph.

[0012] Furthermore, the process of filtering candidate nodes to obtain the corresponding archived knowledge subgraph includes: Obtain multiple candidate node combinations formed between candidate nodes from different unrecorded semantic segments corresponding to all unrecorded semantic segments; The uncollected folk bone-setting prescriptions are decomposed into the smallest semantic units, and the number of smallest semantic units that each candidate node combination can cover is determined. Based on the candidate node combination with the largest coverage, the filtered candidate nodes are determined and the archived knowledge subgraph corresponding to the fuzzy knowledge subgraph is obtained.

[0013] The present invention has the following beneficial effects: This invention constructs a semantic field by establishing core nodes and a basic topological structure based on the TCM standard knowledge system, ensuring the systematic nature and standardized structure of the basic framework layer. Based on the similarity between semantic fragments in already included folk bone-setting prescriptions and core nodes, it establishes mapping and extension relationships between text data and core nodes, obtaining a semantic association path from folk expression to standard knowledge. Furthermore, it utilizes the co-occurrence relationships between included semantic fragments to construct relational edges in the semantic association path and determine their weights, quantifying the strength and reliability of knowledge associations to build the target semantic field, thus achieving the structured integration of unstructured folk knowledge. For new, unincluded folk bone-setting prescriptions, based on deep learning semantic analysis and a graph structure reasoning engine, it achieves accurate disambiguation and knowledge graph generation of fuzzy text through multiple screening of candidate nodes corresponding to unincluded semantic fragments. This fully preserves the original information in the text data, such as prescription compatibility logic and regional colloquialisms, ultimately achieving a precise association between the systematic knowledge of bone-setting prescriptions and the original text data. This provides reliable technical support for the systematic management, accurate inheritance, and promotion of folk bone-setting prescriptions, effectively solving the problems of knowledge fragmentation, value loss, and ambiguity caused by traditional processing methods. Attached Figure Description

[0014] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 A flowchart illustrating the steps of a knowledge service platform for the inheritance and promotion of traditional bone-setting remedies provided in one embodiment of the present invention; Figure 2 A detailed flowchart of step S1 in a knowledge service platform for the inheritance and promotion of folk bone-setting remedies provided in an embodiment of the present invention; Figure 3 A detailed flowchart of step S1 in a knowledge service platform for the inheritance and promotion of folk bone-setting remedies provided in another embodiment of the present invention; Figure 4 A detailed flowchart of step S2 in a knowledge service platform for the inheritance and promotion of folk bone-setting remedies provided in an embodiment of the present invention; Figure 5 A detailed flowchart of step S3 in a knowledge service platform for the inheritance and promotion of folk bone-setting remedies provided in an embodiment of the present invention; Figure 6This is a schematic diagram of the hardware operating environment of a knowledge service device for the inheritance and promotion of folk bone-setting methods, as described in an embodiment of the present invention. Figure 7 This is a schematic diagram of the framework structure of a knowledge service platform for the inheritance and promotion of folk bone-setting methods, which is involved in the embodiments of the present invention. Detailed Implementation

[0016] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a knowledge service platform for the inheritance and promotion of traditional bone-setting remedies proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0018] The following description, in conjunction with the accompanying drawings, details a specific solution for a knowledge service platform provided by the present invention for the inheritance and promotion of traditional bone-setting remedies.

[0019] Example 1: For a knowledge service platform provided by this invention for the inheritance and promotion of traditional bone-setting remedies, please refer to [link / reference]. Figure 7 , Figure 7 This is a schematic diagram of the framework structure of a knowledge service platform for the inheritance and promotion of folk bone-setting methods, which is involved in the embodiments of the present invention.

[0020] The knowledge service platform for the inheritance and promotion of folk bone-setting remedies (hereinafter referred to as the "inheritance and promotion knowledge service platform") includes: The semantic field construction module A10 is used to construct the core nodes of the semantic field and the basic topology between the core nodes based on the standard knowledge system. According to the similarity between the semantic fragments in the collected folk bone-setting prescription texts and the core nodes, the module constructs the mapping relationship between text and nodes in the basic topology and expands the nodes to obtain semantic association paths. Based on the co-occurrence relationship between the collected semantic fragments in the collected folk bone-setting prescription texts, the module constructs relation edges in the semantic association paths and determines their weight values ​​based on the source of the relation edges to obtain the target semantic field. The knowledge archiving module A20 is used to filter candidate nodes to obtain an archived knowledge subgraph based on the set of candidate nodes in the target semantic field of unrecorded semantic fragments in the unrecorded folk bone-setting prescription texts, the relationship edges between the candidate nodes and the weight values ​​of the relationship edges, so as to archive folk bone-setting prescription knowledge.

[0021] Please see Figure 1 , Figure 1 The diagram illustrates a flowchart of the steps involved in the knowledge service platform for the inheritance and promotion of traditional bone-setting remedies provided by an embodiment of the present invention.

[0022] The methods and steps corresponding to the knowledge service platform for the inheritance and promotion of folk bone-setting remedies include: Step S1: Based on the standard knowledge system, construct the core nodes of the semantic field and the basic topological structure between the core nodes. According to the similarity between the semantic fragments in the collected folk bone-setting prescription texts and the core nodes, construct the mapping relationship between the text and the nodes in the basic topological structure and expand the nodes to obtain the semantic association path. In this embodiment, records of folk bone-setting remedies are collected in advance through visits and surveys, including archived materials from primary healthcare institutions, oral records of folk TCM inheritors, local chronicles and medical literature, and research materials from TCM colleges and universities. For the bone-setting prescriptions that have been collected and verified, standardize the processing, organize and supplement the text information of the prescriptions, including: Basic information on bone-setting formulas (formula name, common name, origin region, information on inheritors, scope of indications, etc.), drug composition information (medicinal materials, dosage description, place of origin of medicinal materials, etc.), preparation process information (description of processing method, order of compatibility, special processing techniques, etc.), application method information (route of administration, frequency of administration, course of treatment, contraindications, etc.), and efficacy evaluation information (description of typical cases, statistics of cure rate, etc.); similarly, textual data of new folk bone-setting formulas not included in the database were also obtained; When processing folk bone-setting prescriptions from different knowledge sources, conflicts in folk knowledge are inevitable. For example, the composition of a bone-setting prescription with the same name may differ in different regions, the dosage of the same medicinal material may vary among different inheritors, and the treatment plan for the same type of fracture may differ in different medical records. Therefore, it is necessary to construct an intermediate semantic layer. First, an intermediate semantic layer (i.e., a semantic field) that can accommodate the expression logic of folk knowledge should be established. In this semantic layer, the knowledge is structurally reorganized and then mapped to the standard knowledge system.

[0023] Specifically, step S1, which involves constructing the core nodes of the semantic field and the basic topological structure between the core nodes based on a standard knowledge system, includes: Based on the standard knowledge system of traditional Chinese medicine, each standard knowledge entity in it is used as the corresponding core node in the semantic field. Based on the knowledge types of the core nodes, basic relationship types between the core nodes are constructed, and a basic topology is formed based on the core nodes and basic relationship types.

[0024] In this embodiment, the system (knowledge service platform) pre-imports the standard knowledge system of traditional Chinese medicine as the skeleton layer of the semantic field, including the standard list of medicinal materials in the pharmacopoeia, the compatibility theory framework in formulary, the disease classification system of orthopedics, etc. Each standard knowledge entity serves as a core skeleton node (referred to as a "core node" or "standard node"). The core skeleton node carries its normative attributes within the standard system (normative attributes can correspond to the attribute nodes of the core node in the semantic field). For example, the medicinal material node carries attributes such as standard name, Latin name, properties and meridian tropism, efficacy and indications, usage and dosage; the prescription node carries attributes such as composition structure, compatibility rules, and indications; and the disease node carries attributes such as etiology and pathogenesis, symptoms and signs, and treatment principles. It should be noted that the above-mentioned medicinal material nodes, prescription nodes, and disease nodes can be regarded as knowledge types of core nodes.

[0025] Based on the above knowledge types, the core skeleton nodes are pre-defined with basic relationship types, including the compatibility relationship between medicinal materials, the composition relationship between prescriptions and medicinal materials, and the treatment relationship between prescriptions and diseases. These basic relationship types constitute the basic topological structure of the semantic field.

[0026] Specifically, please refer to Figure 2 Step S1, based on the similarity between semantic fragments and core nodes in the collected folk bone-setting prescription texts, constructs a mapping relationship between texts and nodes in the basic topology, including: Step S11: Use a preset deep learning model to identify the semantic fragments in the collected folk bone-setting prescription texts, and determine the target core node in the basic topology that corresponds to the target collected semantic fragment. Step S12: Determine the cosine similarity between the corresponding vectors of the target semantic fragments and the target core nodes, and establish a mapping relationship between the target semantic fragments and the target core nodes whose cosine similarity is greater than or equal to a preset similarity threshold.

[0027] The system performs deep semantic analysis on the original texts of complete folk bone-setting prescriptions that have been published, included, and verified for efficacy (i.e., the included folk bone-setting prescription texts), extracting the knowledge relationships contained therein. The text analysis and recognition can use a sequence labeling model based on deep learning. This model has been pre-trained on a large number of traditional Chinese medicine literatures and can recognize semantic fragments such as names of medicinal materials, dosage descriptions, processing descriptions, efficacy statements, and disease symptoms in the text. It should be noted that in step S11, the set of candidate core nodes is obtained by retrieving them from the basic topology. Then, the cosine similarity between the target semantic fragment and each node in the set of candidate core nodes is calculated, and the candidate core nodes with similarity greater than the preset similarity threshold are mapped.

[0028] For each identified (included) semantic fragment, the system retrieves its corresponding relevant standard nodes in the above skeleton layer (basic topology). The following explanation uses the target included semantic fragment (referring to any included semantic fragment) and its corresponding target core node (any core node) as an example.

[0029] Using the existing public deep learning language model BERT-TCM, semantic segments and standard nodes are transformed into encoded vectors, and the cosine similarity between the target semantic segments and the target core nodes in the semantic field is calculated. : ; Where s represents the semantic fragment included in the s-th target. Let be the encoding vector of the s-th semantic segment, and v represent the v-th target core node in the skeleton layer. This is the encoding vector for the v-th node; Represents the magnitude of the vector; To avoid dividing the denominator by zero, use extremely small constants (e.g., 0.00001). like If the similarity threshold is greater than or equal to the preset threshold (e.g., 0.7), then the two are considered to be successfully matched, and a mapping relationship from unofficial expressions to standard nodes can be established.

[0030] Specifically, please refer to Figure 3 Step S1, based on the similarity between semantic fragments and core nodes in the collected folk bone-setting prescription texts, constructs a mapping relationship between text and nodes in the basic topology and expands nodes to obtain semantic association paths, including: Step S101: Create extended nodes in the basic topology for target semantic segments with cosine similarity less than a preset similarity threshold. Step S102: Based on the semantic approximation relationship between the extended nodes and the core nodes, construct the connection relationship between the extended nodes and the core nodes, and obtain the semantic association path based on the mapping relationship and the connection relationship.

[0031] Based on the above embodiments, in this embodiment, when no core node corresponding to the semantic segment is found or the cosine similarity is less than the preset similarity threshold, it can be considered that there are folk names or local expressions not included in the skeleton layer in the collected folk bone-setting text. This information will not be discarded, but new semantic nodes will be created in the basic topology. These new nodes are attached to the core node as an extension layer and are called extension nodes. The extended nodes and core nodes are connected to the corresponding core nodes through semantic similarity relationships such as synonymy, near-synonymy, and regional variant relationships, thus forming a semantic association path from the folk expression of traditional bone-setting texts to standard knowledge in the semantic field.

[0032] Step S2: Based on the co-occurrence relationship between the included semantic fragments in the included folk bone-setting prescription texts, construct relation edges in the semantic association path and determine their weight values ​​based on the source of the relation edges to obtain the target semantic field; Specifically, please refer to Figure 4 Step S2, which involves constructing relation edges in the semantic association path based on the co-occurrence relationships between semantic segments in the collected folk bone-setting prescription texts, includes: Step S21: Determine the point-like mutual information representing the co-occurrence relationship between any two semantic fragments in the collected collection of folk bone-setting prescriptions. Step S22: Construct a relationship edge between two included semantic segments whose point-based mutual information is greater than a preset correlation threshold in the semantic association path.

[0033] In this embodiment, based on the collected folk bone-setting prescription text dataset, the co-occurrence relationship between the (collected) semantic fragments is inferred; When two semantic fragments frequently co-occur in multiple texts, a semantic relationship can be inferred between them. The statistical significance of co-occurrence is calculated, and significant co-occurrence patterns are transformed into relational edges in the semantic field. Specifically: Based on a corpus containing all recorded folk bone-setting prescription texts (the corpus of recorded folk bone-setting prescription texts), for any two semantic segments A and B in the text set, calculate their pointwise mutual information (PMI): ; Where N is the total number of texts in the text set. The number of texts containing semantic segment A. The number of texts containing semantic segment B. The number of texts that contain both A and B; To avoid dividing the denominator by zero, use extremely small constants (e.g., 0.00001). when When the co-occurrence of the two is greater than a preset relevance threshold (e.g., 3), it is considered statistically significant, and a relation edge between the two is established in the semantic field. ; Based on the attributes of A and B, the type of relation edge can also be directly identified. The types of relation edges include: the co-occurrence of medicinal materials and processing methods forms processing relation edges; the co-occurrence of medicinal materials and efficacy descriptions forms efficacy relation edges; the co-occurrence of medicinal materials forms compatibility relation edges; and the co-occurrence of medicinal materials and disease symptoms forms indication relation edges. For example, if the system finds that Drynaria fortunei often co-occurs with expressions such as tonifying the kidneys and strengthening bones, processing with black bean juice, and soreness and weakness of the waist and knees, then it establishes corresponding relation edges in the semantic field.

[0034] Specifically, step S2, determining the weight value of a relation edge based on its source, includes: The initial weight value of the relation edge is determined based on the source authority evaluation, the weight increase is determined based on the verification frequency of the relation edge, and the updated weight value of the relation edge is obtained by updating the initial weight value using the weight increase.

[0035] In this embodiment, each relation edge in the semantic field is first assigned an initial weight value, which is between 0 and 1. Based on the authoritative evaluation of the source corresponding to the relation edge (which can be defined manually), relation edges from authoritative literature such as pharmacopoeias, classic medical books, and clinical guidelines are assigned a high credibility coefficient (e.g., 0.7-1); relations from oral traditions and local literature are assigned a medium credibility coefficient (e.g., 0.3-0.6); and relations from a single case record are assigned a low credibility coefficient (e.g., 0-0.2). When a relation edge appears repeatedly from multiple independent sources, its initial weight increases with the number of validations. Therefore, the update function for generating the initial weight is as follows: ; in, This represents the updated weight value. This represents the initial weight value, and n is the number of times the relation edge is validated in different text data sources. The preset saturation verification frequency threshold (e.g., set to 1000) is used, employing a logarithmic function to avoid excessive inflation of the weights of high-frequency relationships; when the actual number of verifications n is greater than... When n is fixed at a certain value, To avoid the result being greater than 1, which could lead to weight overflow.

[0036] The larger the ratio, the higher the frequency of verification of the relation edge, and the higher its reliability and prevalence. The product of the two represents the potential for improvement of the edge weights of the relation. The weight increase is the amount that varies with the number of verifications and the verification frequency, compared to the initial base weight. Add them together to get the updated weights of the corresponding edges; The weights mentioned above reflect the reliability and prevalence of the relation edge. The weight update mechanism enables the semantic field to distinguish between core knowledge and peripheral knowledge, and prioritizes the use of high-weight paths for reasoning during knowledge transformation.

[0037] Based on the implementation process of the above embodiments, a constructed target semantic field is obtained. A completed semantic field generally includes: The core skeleton nodes include medicinal material nodes, prescription nodes, disease nodes, etc.; the core skeleton nodes are interconnected; and each core skeleton node is associated with all its attribute nodes, mapping relationships, and extended nodes in the semantic field.

[0038] For example, the medicinal herb elderberry is recorded in the semantic field not only by its scientific name Sambucus williamsii, but also by its folk names (such as Seven-leaf Lotus, Bone-setting Pill, etc.), morphological characteristics (such as opposite leaves, white pith, etc.), traditional efficacy (such as tendon and bone healing, blood circulation and pain relief, etc.), common medicinal combinations (such as being used with Drynaria fortunei and natural copper), and regional distribution characteristics (such as a certain variety being used more often in North China). All existing publicly available bone-setting formulas can be converted into semantic field formats using the methods described above; Then, when extracting text data containing information such as regional information, common names, and efficacy from the original text of folk bone-setting prescriptions, whether it is about medicinal materials, prescriptions, or pathology, it can be matched in the semantic field. Even if there are differences in individual names, combinations, descriptions, etc., it is possible to accurately understand the medicinal materials, prescriptions, and pathological entities described in the text by matching multiple relational paths in the semantic field.

[0039] Step S3: Based on the set of candidate nodes in the target semantic field of unrecorded semantic fragments in the unrecorded folk bone-setting prescription texts, the relationship edges between the candidate nodes and the weight values ​​of the relationship edges, the candidate nodes are screened to obtain the archived knowledge subgraph for archiving folk bone-setting prescription knowledge.

[0040] In this embodiment, constructing a semantic field solves the problem of identifying and mapping elements of folk knowledge. However, the true essence of folk bone-setting remedies lies in their holistic nature and context-dependent nature. A prescription (i.e., a bone-setting remedy) is not simply a list of medicinal materials; its efficacy stems from the comprehensive effect of the materials within a specific context of combination, processing, and treatment of a specific ailment. Traditional data processing methods decompose prescriptions into independent fields (materials, dosage, processing), which precisely disrupts this inherent, networked knowledge structure. Furthermore, the unstructured nature of folk bone-setting remedies stems from the cognitive gap between their experiential expression system and standardized knowledge system. The fundamental flaw of traditional item-by-item standardization methods lies in severing the inherent integrity and contextual dependence of folk knowledge. Therefore, this embodiment does not choose to directly process newly entered bone-setting remedies item by item.

[0041] Therefore, this embodiment does not treat the input folk prescription text as a document from which discrete fields need to be extracted, but rather as a fuzzy knowledge image segment to be parsed. The task of the knowledge parsing engine constructed in this embodiment is to find the most reasonable and complete "standardized knowledge graph" interpretation for this fuzzy segment in the constructed global semantic field.

[0042] Specifically, please refer to Figure 5 Step S3, based on the set of candidate nodes in the target semantic field of unrecorded semantic fragments from unrecorded folk bone-setting prescription texts, the relationship edges between candidate nodes, and the weight values ​​of the relationship edges, filters candidate nodes to obtain an archived knowledge subgraph, including: Step S31: Identify the target unrecorded semantic fragment in the unrecorded folk bone-setting prescription text, and retrieve each candidate node corresponding to the target unrecorded semantic fragment in the target semantic field to obtain a candidate node set; Step S32: Based on the candidate node set and the relationship edges and weights of the candidate nodes, a fuzzy knowledge subgraph of uncollected folk bone-setting prescription texts is obtained, and candidate nodes are filtered to obtain the corresponding archived knowledge subgraph.

[0043] In this embodiment, when a new, unrecorded original text describing a bone-setting remedy is input (unrecorded folk bone-setting remedies), such as: Three qian of Sichuan teasel root and five qian of dried bone floss stir-fried with black bean juice are decocted together with elderberry. This is used for the later stage of femoral neck fracture in the elderly. Yellow wine can be added as a guide. The system's deep learning semantic analysis and graph structure reasoning engine first performs deep semantic analysis: using the sequence labeling model trained in the above embodiments, it identifies all semantic segments in the text, such as: Sichuan teasel root, three qian, drynaria rhizome, stir-fried with black bean juice, five qian, elderberry, decocted together, late stage of femoral neck fracture in the elderly, and yellow wine as a guide. Obtain a set of semantic fragments ; For each of the identified semantic segments above, the target unrecorded semantic segment is... For example, the engine retrieves all possible corresponding nodes (not limited to standard nodes, attribute nodes, and extended nodes) in the established semantic field and generates a candidate node set; For example, "stir-frying with black bean juice" will activate the "roasting with black bean juice" standard node under the "processing method" node, and may also activate a process extension node describing "stir-frying with black bean juice". In the semantic field, obtain the possible relationship edges between all these candidate nodes, obtain the weight value of the relationship edge and label the specific relationship edge type, including processing relationship edge, efficacy relationship edge, compatibility relationship edge, indication relationship edge; Ultimately, the engine transforms the text into a fuzzy knowledge subgraph with multiple candidates, referred to here as a fuzzy knowledge subgraph; it contains several semantic segments, each of which contains several candidate nodes; The nodes and relation edges in this fuzzy knowledge subgraph are all uncertain (because there are multiple candidates). It is a graphical representation of the original text semantic structure, but it has not yet been disambiguated.

[0044] The fuzzy knowledge subgraph generated in the previous step is chaotic and may be self-contradictory. Each semantic fragment in the fuzzy knowledge subgraph corresponds to multiple candidate nodes, forming an ambiguous candidate space. The essence of disambiguation is to make a series of choices in this candidate space to filter the candidate nodes, and finally determine a unique semantic field node for each semantic fragment to obtain a knowledge subgraph that can be archived.

[0045] More specifically, step S32, filtering candidate nodes to obtain the corresponding archived knowledge subgraph, includes: Candidate nodes are filtered using preset constraint rules to obtain the archived knowledge subgraph corresponding to the fuzzy knowledge subgraph. The preset constraint rules include at least one or more of the following: The first constraint rule is that there is a direct relationship edge between adjacent candidate nodes in the target semantic field; the second constraint rule is that the node types corresponding to the candidate nodes have a necessary relational dependency. The first and second constraint rules are used to filter out candidate nodes in the candidate node set that do not meet the rules.

[0046] In this embodiment, a constraint propagation network is used to transform the fuzzy knowledge subgraph. This constraint network can eliminate obviously unreasonable candidate nodes through two types of preset constraint rules, thereby narrowing the disambiguation space. Constraint Rule 1: Candidate nodes of adjacent semantic segments in the input text (excluding folk bone-setting prescriptions) must have a direct relation edge in the semantic field; Example: The sentence "five qian of Drynaria fortunei after being stir-fried with black bean juice" in the original text can be broken down into two semantic segments: "Drynaria fortunei" and "stir-fried with black bean juice". For candidate nodes of "bone-strengthening herb" and "stir-fried with black bean juice", the semantic field must contain the term "medicinal herb". The direct connection between "processing methods" and "preparation methods"; If a candidate for "Drynaria fortunei" (e.g., misidentified as "Drynaria fortunei") has no processing relationship with "stir-fried with black bean juice", then this candidate for "Drynaria fortunei" will be directly removed.

[0047] Constraint Rule 2: The node types of candidate nodes have necessary relational dependencies; For example, the type of prescription node must have at least one compositional relationship edge pointing to the type of medicinal material node; Check whether the existing constituent medicinal materials of the candidate prescription node have knowledge overlap with the candidate medicinal material nodes described in the current text. If there is no overlap, the candidate prescription node is removed. After filtering using two types of constraints, the number of candidate nodes for each semantic segment is significantly reduced, and the ambiguity of the fuzzy knowledge subgraph is significantly reduced.

[0048] This embodiment can be used as the first step in screening candidate nodes.

[0049] Further, step S32, filtering candidate nodes to obtain the corresponding archived knowledge subgraph, includes: Determine the target relation path formed by relation edges between any candidate nodes corresponding to any two unrecorded semantic segments; By using the weights and values ​​of all relation edges on the target relation path, the path strength of the target relation path is determined, and the connection strength between the corresponding candidate nodes is obtained. Based on the connection strength, the connectivity of the target candidate node is determined, and the candidate node with the lowest connectivity is iteratively eliminated to obtain the filtered candidate nodes and the archived knowledge subgraph corresponding to the fuzzy knowledge subgraph.

[0050] It should be noted that when iteratively eliminating candidate nodes with the lowest connectivity, if multiple candidate nodes have the same lowest connectivity, the node with the lowest semantic similarity to the text context is eliminated first, or one is randomly eliminated to prevent the algorithm from going into an infinite loop.

[0051] Based on the above embodiments, in this embodiment, the fuzzy knowledge subgraph has m semantic segments. Each semantic fragment have One candidate node; Then, for any two semantic fragments in the fuzzy knowledge subgraph , The system searches for relational paths connecting all candidate node pairs within the semantic field. The path search employs a breadth-first strategy, limiting the maximum path length to 3 hops (meaning the number of relational edges between nodes in the knowledge graph does not exceed 3) to avoid introducing overly indirect relationships. For each candidate node pair (a, b), where a is a semantic fragment... Candidate b is a semantic fragment The system records all paths within 3 hops from a to b in the semantic field; The target relationship path found between a and b The weights of all relation edges on the path are summed to obtain the weight sum value, which is then divided by the number of relation edges to obtain the path strength. The path strength reflects the overall reliability of all relations on the path. For each pair of candidate nodes (a, b), if there are multiple connection paths, the maximum value of all path strengths is taken as the connection strength of the candidate node pair (a, b); thus, the connection strengths of all pairs of candidate nodes (a, b) are obtained. For semantic fragments For a given candidate node a, the connectivity of the a-th candidate node is obtained by summing the connection strengths of all candidate node pairs connected to a. Then, the candidate nodes with the lowest connectivity are iteratively eliminated. After each round of iteration, the connectivity of all remaining candidate nodes is recalculated until a semantic fragment is reached. The process continues until the connection strength of all remaining candidate nodes exceeds a preset strength threshold (e.g., set to 60% of the maximum connection strength) or until only one candidate node remains. This embodiment can be used as the second step in screening candidate nodes.

[0052] Further, step S32, filtering candidate nodes to obtain the corresponding archived knowledge subgraph, includes: Obtain multiple candidate node combinations formed between candidate nodes from different unrecorded semantic segments corresponding to all unrecorded semantic segments; The uncollected folk bone-setting prescriptions are decomposed into the smallest semantic units, and the number of smallest semantic units that each candidate node combination can cover is determined. Based on the candidate node combination with the largest coverage, the filtered candidate nodes are determined and the archived knowledge subgraph corresponding to the fuzzy knowledge subgraph is obtained.

[0053] Based on the above embodiments, in this embodiment, if each semantic segment in the fuzzy knowledge subgraph may still have several candidate nodes remaining; Then all (unrecorded) semantic segments of the fuzzy knowledge subgraph will have several different combinations of candidate nodes, and each combination represents a complete interpretation scheme for the entire fuzzy knowledge subgraph; For example, assuming semantic fragments There are two remaining candidate nodes, a1 and a2, and semantic fragments. There are two remaining candidate nodes, b1 and b2, with semantic fragments. There is one candidate node c1 remaining; The possible complete explanations (combinations) are as follows: Combination 1: Select a1, b1, c1; Combination 2: Choose a1, b2, c1; Combination 3: Choose a2, b1, c1; Combination 4: Choose a2, b2, c1; At this point, the system decomposes the original text into the smallest semantic units, which are the sets of content words obtained after fine-grained word segmentation of the original text. Suppose the original text has several minimal semantic units: (Dipsacus asper, three qian, Drynaria fortunei, stir-fried with black bean juice, five qian, elderberry, decocted together, used for, elderly, femoral neck fracture, later stage, can be added, yellow wine, as a guide); These are divided into content words (core information with actual clinical significance) and function words (words that connect or modify, such as "together to be decocted", "can be added", and "as a guide"). Check how many minimum semantic units each combination's knowledge subgraph can cover, that is, determine the number of minimum semantic units that each candidate node combination can cover; Prioritize selecting the combination that covers the largest number of the smallest semantic units; When the coverage quantity is the same, choose the combination that covers more content words.

[0054] Finally, for each variable (semantic fragment) in the fuzzy knowledge subgraph, a unique candidate node is selected after filtering, eliminating all ambiguity and generating a definite and refined archived knowledge subgraph; For example, the output diagram will clearly show newly entered bone-setting prescriptions: This is a prescription node named (for example, an automatically generated temporary ID), whose components include Dipsacus asper (dosage: 3 qian), Drynaria fortunei (processed: roasted with black bean juice, dosage: 5 qian), and Sambucus chinensis; its preparation process is decoction; its main indication is femoral neck fracture (stage: late stage, population: elderly); its excipient is yellow rice wine as a guide.

[0055] This embodiment, through the construction of an intermediate semantic layer, can accurately disambiguate and generate knowledge graphs for unstructured bone-setting prescription texts, fully preserving original information such as prescription compatibility logic and regional colloquial names, and standardizing the inclusion of newly entered bone-setting prescriptions.

[0056] Furthermore, to facilitate traceability and retrieval, a unique archiving identifier (associated with the original data ID from the collection phase) can be assigned to each knowledge subgraph. The identifier embeds core information codes (such as "region-fracture type-prescription type") to achieve bidirectional traceability between the knowledge graph and the original collected data, ensuring that archived knowledge can be linked back to original information such as inheritors and collection scenarios. A retrieval index is built based on the core attributes of the archived knowledge graph (prescription name, medicinal material composition, processing technology, indications, geographical origin, and credibility). The index is associated with the complete information of the knowledge subgraph and the original text fields, and supports fast retrieval by a single attribute (such as "bone-setting prescription containing Drynaria fortunei") or a combination of multiple attributes (such as "North China region - external application - limb fractures - high credibility"), providing efficient support for knowledge promotion and clinical query.

[0057] This invention constructs a semantic field by establishing core nodes and a basic topological structure based on the TCM standard knowledge system, ensuring the systematic nature and standardized structure of the basic framework layer. Based on the similarity between semantic fragments in already included folk bone-setting prescriptions and core nodes, it establishes mapping and extension relationships between text data and core nodes, obtaining a semantic association path from folk expression to standard knowledge. Furthermore, it utilizes the co-occurrence relationships between included semantic fragments to construct relational edges in the semantic association path and determine their weights, quantifying the strength and reliability of knowledge associations to build the target semantic field, thus achieving the structured integration of unstructured folk knowledge. For new, unincluded folk bone-setting prescriptions, based on deep learning semantic analysis and a graph structure reasoning engine, it achieves accurate disambiguation and knowledge graph generation of fuzzy text through multiple screening of candidate nodes corresponding to unincluded semantic fragments. This fully preserves the original information in the text data, such as prescription compatibility logic and regional colloquialisms, ultimately achieving a precise association between the systematic knowledge of bone-setting prescriptions and the original text data. This provides reliable technical support for the systematic management, accurate inheritance, and promotion of folk bone-setting prescriptions, effectively solving the problems of knowledge fragmentation, value loss, and ambiguity caused by traditional processing methods.

[0058] Example 2: This invention also proposes a knowledge service device for the inheritance and promotion of traditional bone-setting remedies. The device can be a computer, a server, or a combination of multiple devices.

[0059] like Figure 6 As shown, Figure 6 This is a schematic diagram of the hardware operating environment of a knowledge service device for the inheritance and promotion of folk bone-setting methods, which is involved in the embodiments of the present invention.

[0060] like Figure 6 As shown, the knowledge service device for the inheritance and promotion of traditional bone-setting remedies may include: a processor 1001, such as a CPU; a network interface 1004; a user interface 1003; a memory 1005; and a communication bus 1002. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display or an input unit such as a control panel; optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory 1005 may be a high-speed RAM or a stable, non-volatile memory, such as a disk drive. Optionally, the memory 1005 may also be a storage device independent of the aforementioned processor 1001. The memory 1005, as a computer storage medium, may include a knowledge service program for the inheritance and promotion of traditional bone-setting remedies (hereinafter referred to as the "inheritance and promotion knowledge service program").

[0061] Those skilled in the art will understand that Figure 6 The hardware structure shown does not constitute a limitation on the device and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0062] Continue to refer to Figure 6 , Figure 6 The memory 1005, which is a computer-readable storage medium, may include an operating device, a user interface module, a network communication module, and a knowledge service program for the inheritance and promotion of folk bone-setting remedies.

[0063] exist Figure 6 In this embodiment, the network communication module is mainly used to connect to the server and can communicate with the server for data; while the processor 1001 can call the knowledge service program stored in the memory 1005 for the inheritance and promotion of folk bone-setting methods and execute the steps in the above embodiments.

[0064] Based on the hardware structure of the knowledge service device for the inheritance and promotion of folk bone-setting remedies described above, various embodiments of the knowledge service platform for the inheritance and promotion of folk bone-setting remedies of the present invention are implemented.

[0065] Furthermore, the present invention also provides a computer-readable storage medium. The computer-readable storage medium stores a knowledge service program for the inheritance and promotion of traditional bone-setting remedies, wherein, when executed by a processor, the knowledge service program for the inheritance and promotion of traditional bone-setting remedies implements the steps of the aforementioned method corresponding to the knowledge service platform for the inheritance and promotion of traditional bone-setting remedies.

[0066] The method implemented when the knowledge service program for the inheritance and promotion of folk bone-setting remedies is executed can be referred to in various embodiments of the knowledge service platform for the inheritance and promotion of folk bone-setting remedies of this invention, and will not be repeated here.

[0067] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0068] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0069] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0070] The above description is only a preferred embodiment of the present invention and does not limit the scope of protection of the present invention. All equivalent structural / method transformations made under the inventive concept of the present invention using the contents of the present invention specification and drawings, or direct / indirect applications in other related technical fields, are included within the scope of protection of the present invention.

Claims

1. A knowledge service platform for the inheritance and promotion of traditional folk bone-setting remedies, characterized in that, The platform includes: The semantic field construction module is used to construct the core nodes of the semantic field and the basic topology between the core nodes based on the standard knowledge system. According to the similarity between the semantic fragments in the collected folk bone-setting prescription texts and the core nodes, the module constructs the mapping relationship between text and nodes in the basic topology and expands the nodes to obtain semantic association paths. Based on the co-occurrence relationship between the collected semantic fragments in the collected folk bone-setting prescription texts, the module constructs relation edges in the semantic association paths and determines their weight values ​​based on the source of the relation edges to obtain the target semantic field. The knowledge archiving module is used to filter candidate nodes in the target semantic field based on the set of candidate nodes in the unrecorded semantic fragments in the unrecorded folk bone-setting prescription texts, the relationship edges between the candidate nodes and the weight values ​​of the relationship edges, to obtain an archived knowledge subgraph for archiving folk bone-setting prescription knowledge.

2. The knowledge service platform for the inheritance and promotion of folk bone-setting remedies according to claim 1, characterized in that, The core nodes of the semantic field constructed based on the standard knowledge system and the basic topological structure between the core nodes include: Based on the standard knowledge system of traditional Chinese medicine, each standard knowledge entity in it is used as the corresponding core node in the semantic field. Based on the knowledge types of the core nodes, basic relationship types between the core nodes are constructed, and a basic topology is formed based on the core nodes and basic relationship types.

3. The knowledge service platform for the inheritance and promotion of folk bone-setting remedies according to claim 1, characterized in that, Based on the similarity between semantic fragments and core nodes in the collected texts of folk bone-setting remedies, a mapping relationship between text and nodes is constructed in the basic topology, including: Using a pre-defined deep learning model, we can identify the semantic fragments included in the existing folk bone-setting prescriptions texts and determine the target core nodes in the basic topology that correspond to the target semantic fragments. Determine the cosine similarity between the corresponding vectors of the target semantic segments and the target core nodes, and establish a mapping relationship between the target semantic segments and the target core nodes whose cosine similarity is greater than or equal to a preset similarity threshold.

4. The knowledge service platform for the inheritance and promotion of folk bone-setting remedies according to claim 3, characterized in that, The process involves constructing a mapping relationship between texts and nodes in the basic topology and expanding nodes to obtain semantic association paths based on the similarity between semantic fragments and core nodes in the collected folk bone-setting prescription texts. This includes: Semantic fragments with a cosine similarity less than a preset similarity threshold are created as extended nodes in the basic topology. Based on the semantic approximation relationship between the extended nodes and the core nodes, the connection relationship between the extended nodes and the core nodes is constructed, and the semantic association path is obtained based on the mapping relationship and the connection relationship.

5. The knowledge service platform for the inheritance and promotion of folk bone-setting remedies according to claim 1, characterized in that, Based on the co-occurrence relationships between included semantic fragments in the already included folk bone-setting prescription texts, relation edges are constructed in the semantic association path, including: Determine the point-like mutual information representing the co-occurrence relationship between any two semantic fragments in the collected collection of folk bone-setting prescriptions; For two included semantic fragments whose point-based mutual information is greater than a preset relevance threshold, a relationship edge is constructed between them in the semantic association path.

6. The knowledge service platform for the inheritance and promotion of folk bone-setting remedies according to claim 1, characterized in that, The weight value of a relation edge is determined based on its source, including: The initial weight value of the relation edge is determined based on the source authority evaluation, the weight increase is determined based on the verification frequency of the relation edge, and the updated weight value of the relation edge is obtained by updating the initial weight value using the weight increase.

7. The knowledge service platform for the inheritance and promotion of folk bone-setting remedies according to claim 1, characterized in that, Based on the candidate node set in the target semantic field of unrecorded semantic fragments from uncollected folk bone-setting prescription texts, and the relationship edges and weights between these candidate nodes, an archived knowledge subgraph is obtained by filtering candidate nodes, including: Identify the target unrecorded semantic fragments in the texts of unrecorded folk bone-setting remedies, and retrieve the candidate nodes corresponding to the target unrecorded semantic fragments in the target semantic field to obtain a set of candidate nodes; Based on the candidate node set and the relationship edges between the candidate nodes and their weights, a fuzzy knowledge subgraph of uncollected folk bone-setting prescriptions is obtained. Candidate nodes are then filtered to obtain the corresponding archived knowledge subgraph.

8. The knowledge service platform for the inheritance and promotion of folk bone-setting remedies according to claim 7, characterized in that, The process of filtering candidate nodes yields a corresponding archived knowledge subgraph, including: Candidate nodes are filtered using preset constraint rules to obtain the archived knowledge subgraph corresponding to the fuzzy knowledge subgraph. The preset constraint rules include at least one or more of the following: The first constraint rule is that there is a direct relationship edge between adjacent candidate nodes in the target semantic field; the second constraint rule is that the node types corresponding to the candidate nodes have a necessary relational dependency. The first and second constraint rules are used to filter out candidate nodes in the candidate node set that do not meet the rules.

9. The knowledge service platform for the inheritance and promotion of folk bone-setting remedies according to claim 1, characterized in that, The process of filtering candidate nodes yields a corresponding archived knowledge subgraph, including: Determine the target relation path formed by relation edges between any candidate nodes corresponding to any two unrecorded semantic segments; By using the weights and values ​​of all relation edges on the target relation path, the path strength of the target relation path is determined, and the connection strength between the corresponding candidate nodes is obtained. Based on the connection strength, the connectivity of the target candidate node is determined, and the candidate node with the lowest connectivity is iteratively eliminated to obtain the filtered candidate nodes and the archived knowledge subgraph corresponding to the fuzzy knowledge subgraph.

10. The knowledge service platform for the inheritance and promotion of folk bone-setting remedies according to claim 1, characterized in that, The process of filtering candidate nodes yields a corresponding archived knowledge subgraph, including: Obtain multiple candidate node combinations formed between candidate nodes from different unrecorded semantic segments corresponding to all unrecorded semantic segments; The uncollected folk bone-setting prescriptions are decomposed into the smallest semantic units, and the number of smallest semantic units that each candidate node combination can cover is determined. Based on the candidate node combination with the largest coverage, the filtered candidate nodes are determined and the archived knowledge subgraph corresponding to the fuzzy knowledge subgraph is obtained.

Images