Industrial injury condition auxiliary analysis method and system based on knowledge graph

Abstract

The invention discloses an industrial injury condition auxiliary analysis method and system based on a knowledge graph, and relates to the technical field of knowledge graphs. The method comprises the following steps: acquiring structural information related to the industrial injury condition and to-be-identified injury condition information; constructing an industrial injury knowledge graph according to the structural information related to the industrial injury condition, and embedding the constructed industrial injury knowledge graph into the industrial injury condition analysis model; and performing dynamic injury condition reasoning according to the to-be-identified injury condition information and the path matching condition of the industrial injury knowledge graph by utilizing an industrial injury condition analysis model with the industrial injury knowledge graph to obtain a disability grade identification result and an interpretable reasoning path. According to the invention, an efficient, accurate and extensible intelligent reference scheme can be provided for industrial injury identification.

Description

Technical Field

[0001] This invention relates to the field of knowledge graph technology, and in particular to a method and system for auxiliary analysis of work-related injuries based on knowledge graphs. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] In the field of work injury assessment, the traditional manual approach has long relied on assessors meticulously reading medical records, imaging reports, and surgical records word by word, converting unstructured text into internal understanding before comparing it with paper or electronic assessment clauses at a string-level search level. This process requires repeated confirmation of more than ten core indicators, such as the location of the injury, the value of functional impairment, and the degree of medical dependence. The absence or ambiguity of any indicator triggers secondary file retrieval, supplementary examinations, or expert review, leading to an exponential increase in the average case processing time. More importantly, the assessment clauses contain numerous numerical ranges and degree adverbs, and different assessors exhibit significant individual differences in the probability mapping range for the same adverb, resulting in a high proportion of grade discrepancies during the review process. When the injury involves multi-system complications or rare sites, the scope of manual retrieval is forced to expand to cross-chapter clauses, further increasing the probability of missed cases. While machine learning paradigms map text to a continuous semantic space via deep neural networks and directly output grade probabilities in an end-to-end manner, the labeled samples they rely on must be reviewed line by line by experienced physicians. Labeling costs are rigidly correlated with quality. Furthermore, when the input text contains anatomical locative terms not covered in the training set or new surgical names, the model's activation vectors deviate from the training manifold, causing a sharp drop in confidence and necessitating a rollback for manual review. In addition, the weight matrix of the softmax layer in the classifier cannot correspond to any medical concepts, resulting in an untraceable decision-making path and failing to meet the rigid requirements of interpretable chains of evidence in administrative and judicial processes.

[0004] Currently, existing technologies utilize knowledge graph-driven injury analysis systems to address the aforementioned issues. The assessment clauses no longer exist as natural language paragraphs but are parsed into condition-conclusion rule subgraphs. When a new case enters the system, the natural language processing module first invokes a character-word-entity joint sequence annotation model fine-tuned from the medical corpus to complete entity recognition and relation extraction from the medical record text; subsequently, a subgraph matching algorithm is used to perform isomorphism detection between the case instance graph and the rule subgraph, and a cost-based optimal matching search determines the candidate clause set.

[0005] The system adopts an incremental graph compilation strategy in the data update dimension: when the identification criteria are revised numerically or new injury types are added, the corresponding nodes and edges only need to be inserted or modified in a transactional manner, without retraining the global model; at the same time, the system provides a version control-based backtracking interface to ensure the comparability and consistency of historical cases during the transition between old and new rules.

[0006] However, existing technologies have the following drawbacks in implementing the above-mentioned injury analysis: conflict resolution relies on hard-coded priority numbers, lacks an interpretable source of weights, and when multiple rules of the same priority are applied simultaneously, only random results are returned, failing to provide traceable decision-making basis. The update mechanism uses full script replacement, lacks version isolation, and historical cases cannot reproduce the original conclusions during the period when old and new rules are mixed, seriously affecting the credibility of administrative reconsideration and judicial review. Summary of the Invention

[0007] To address the shortcomings of existing technologies, the purpose of this invention is to provide a knowledge graph-based auxiliary analysis method and system for work-related injury conditions, which can provide an efficient, accurate, and scalable intelligent reference solution for work-related injury assessment.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solution: The first aspect of this invention provides a knowledge graph-based auxiliary analysis method for work-related injury conditions, comprising the following steps: Obtain structured information related to work-related injuries and information on injuries to be assessed; A work injury knowledge graph is constructed based on structured information related to work injury conditions, and the constructed work injury knowledge graph is embedded into the work injury condition analysis model. A work injury analysis model with a work injury knowledge graph is used to perform dynamic injury reasoning based on the path matching between the injury information to be identified and the work injury knowledge graph, so as to obtain the disability level assessment result and the interpretable reasoning path.

[0009] Furthermore, the structured information related to work-related injuries includes the relevant standard clause number, applicable injury clauses, disability rating settings, and associated rules.

[0010] Furthermore, the specific steps for constructing a work injury knowledge graph based on structured information related to work injury are as follows: Knowledge extraction from structured information related to work-related injuries; By integrating knowledge based on the relationships between different pieces of knowledge, a knowledge-related network can be constructed. The knowledge association network is stored in a graph database in the form of a graph structure.

[0011] Furthermore, in the graph structure, structured information related to work-related injuries is used as nodes, and the relationships between these structured information items are used as edges.

[0012] Furthermore, the specific steps for using a work injury severity analysis model with a work injury knowledge graph to perform dynamic injury reasoning based on the path matching between the injury information to be assessed and the work injury knowledge graph are as follows: The injury information to be assessed is preprocessed to obtain standardized injury data; Based on standardized injury data, path matching of the work injury knowledge graph is performed to obtain the reasoning path and reasoning result; The system checks whether the injury information to be identified meets the special rules, and makes inferences and corrections based on the results of the special rule detection.

[0013] Furthermore, the preprocessing steps for information regarding injuries to be assessed include: Analyze the injury information to be identified; Based on the analysis results, the injury information to be identified is mapped to standard terminology.

[0014] Furthermore, when performing path matching for a work injury knowledge graph based on standardized injury data, if multiple paths are matched for an injury, the path superposition effect is determined according to priority settings and anatomical weights.

[0015] A second aspect of this invention provides a knowledge graph-based auxiliary analysis system for work-related injuries, comprising: The data acquisition module is configured to acquire structured information related to work-related injuries and information on injuries to be assessed. The knowledge graph construction module is configured to build a work injury knowledge graph based on structured information related to work injury conditions, and embed the constructed work injury knowledge graph into the work injury condition analysis model. The injury reasoning module is configured to use a work injury analysis model with a work injury knowledge graph to perform dynamic injury reasoning based on the path matching between the injury information to be identified and the work injury knowledge graph, and obtain the disability level assessment result and an interpretable reasoning path.

[0016] A third aspect of the present invention provides a computer-readable storage medium storing a computer program adapted for loading by a processor and executing steps in the knowledge graph-based auxiliary analysis method for work-related injuries as described in the first aspect of the present invention.

[0017] A fourth aspect of the present invention provides a computer device comprising: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the knowledge graph-based auxiliary analysis method for work-related injuries as described in the first aspect of the present invention.

[0018] The above one or more technical solutions have the following beneficial effects: This invention discloses a knowledge graph-based method and system for auxiliary analysis of work-related injuries. Through a semantic alignment mechanism of the graph structure, the logical clauses are transformed into executable paths, freeing injury assessment from manual string comparison and reducing fluctuations in severity levels caused by subjective differences. The graph traversal records completely preserve the semantic identifiers of nodes and edges, allowing the decision-making process to be directly mapped to a traceable chain of evidence, enhancing the transparency of conclusions. Path overlay and conflict resolution strategies automatically arbitrate priorities when multiple clauses are activated simultaneously, reducing conclusion drift caused by rule overlap. Incremental update capabilities allow nodes and relationships to be dynamically replaced at the transaction level, synchronizing with the latest medical or legal changes without service interruption, significantly improving system scalability and continuous service capabilities.

[0019] This invention utilizes a multi-level matching engine and conflict resolution mechanism of knowledge graphs to significantly improve the matching accuracy between injury characteristics and assessment clauses, reduce omissions and subjective biases in manual comparisons, and improve the efficiency of single-case assessments. By visualizing the knowledge graph traversal trajectory, it outputs structured assessment results and transparent reasoning paths, enhancing the interpretability of decisions and meeting the requirements of compliance and audit traceability in work injury assessments.

[0020] This invention also features a dynamic knowledge update interface and incremental expansion capabilities, supporting data synchronization for regulatory revisions, new injury types, or medical knowledge. It effectively addresses the core pain points of low efficiency in traditional manual methods and weak adaptability of machine learning models, providing an efficient, accurate, and scalable intelligent solution for the field of work injury assessment.

[0021] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of this application 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 this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart of the knowledge graph-based auxiliary analysis method for work-related injuries in Embodiment 1 of the present invention; Figure 2 This is a system architecture diagram of the knowledge graph-based auxiliary analysis of work-related injuries in Embodiment 1 of the present invention. Detailed Implementation

[0024] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, 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.

[0025] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof. The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0026] Example 1: Embodiment 1 of this invention provides a knowledge graph-based auxiliary analysis method for work-related injury conditions, such as... Figure 1 and Figure 2 As shown, it includes the following steps: S1: Obtain structured information related to work-related injuries and information on injuries to be assessed.

[0027] In one specific implementation, the structured information related to the work-related injury includes the clause number of the relevant standard, the applicable injury clause, the disability rating, and the association rules. The relevant standard refers to recognized documents in the field, such as the "Standards for the Assessment of Disability Caused by Work-Related Injuries and Occupational Diseases." Information regarding the injury to be assessed includes physical examination reports and auxiliary examination results.

[0028] S2: Construct a work injury knowledge graph based on structured information related to work injury conditions, and embed the constructed work injury knowledge graph into the work injury condition analysis model.

[0029] S2.1: Construct a work injury knowledge graph based on structured information related to work injury conditions.

[0030] S2.1.1: Extract knowledge from structured information related to work-related injuries.

[0031] In one specific implementation, knowledge extraction is performed using a knowledge extraction module. First, terminology is standardized, and duplicate terms are disambiguated and merged. Then, knowledge extraction is performed using triples, which include entities, relations, and attributes.

[0032] Entity extraction: Identify injury type (e.g., "fracture" or "nerve injury"), disability level (e.g., "Level 1" or "Level 2"), and assessment methods (e.g., "joint range of motion measurement" or "muscle strength score").

[0033] Relationship extraction: Extracting the relationships between entities, such as "fracture → leading to → knee joint dysfunction" and "muscle strength score → impact → disability level".

[0034] Attribute extraction: Add descriptive information to entities, such as the attributes of "fracture" including "location", "healing time" and "complications".

[0035] S2.1.2: Integrate knowledge based on the relationships between knowledge points to construct a knowledge association network.

[0036] In one specific implementation, the association network of this embodiment consists of four types of core nodes, corresponding to four dimensions: clauses, injuries, severity levels, and rules. Clause nodes refer to specific clauses in legal texts, policy regulations, or industry standards. Injury nodes refer to medical classifications of injuries. Severity nodes refer to the grading standards for the severity of injuries. Rule nodes refer to logical rules derived from clauses, injuries, and severity levels, such as "If the injury is a skull fracture and the severity level is serious injury level two, then the compensation standard of clause X applies."

[0037] Then, dynamic associations are formed by establishing relationship types between nodes. The association between clause nodes and injury points is determined by the scope of injury specified in the clauses. The association between injury nodes and severity nodes is determined by the correspondence between injury and severity. The association between severity nodes and clause nodes is determined by the applicability of clauses triggered by severity levels. The association between rule nodes and all other nodes is determined by the constraints or derivations of rules on clauses, injuries, or severity levels. Cross-dimensional associations are also established; for example, a clause may be directly associated with a rule, such as "Clause X requires the compensation level to be determined according to Rule Y," or an injury may be indirectly associated with a clause through a rule, such as "Injury A is mapped to Clause X through Rule Y." Then, semantic text recognition is performed based on the association network to solve the problem of different expressions of the same entity. Specifically, the following steps are included: 1. Semantic similarity calculation Semantic similarity calculation is achieved using pre-trained models in the medical field, such as BERT and BioBERT. The training text is encoded into high-dimensional vectors, and cosine similarity is calculated. Word vectors are generated using Word2Vec or GloVe, and auxiliary metrics such as edit distance and Jaccard similarity are used to address differences in short texts.

[0038] 2. Domain ontology mapping Construct a joint ontology of medical and legal terms and perform rule mapping. Integrate ICD-10 (International Classification of Diseases), SNOMED CT (Medical Terminology System), and injury descriptions from the "Regulations on Work Injury Insurance" to establish a standardized terminology database. If an input term has a synonym or hyponym in the ontology database (e.g., "shoulder fracture" is a hyponym of "upper limb fracture"), it will be automatically replaced with the standard term.

[0039] 3. Context-aware disambiguation Dependency parsing and co-occurrence network analysis were performed on the text. Dependency parsing analyzed the grammatical roles of terms in sentences (such as subject and object) and distinguished subtle differences by combining modifiers (such as "mild" and "complete"). Co-occurrence network analysis statistically analyzed the co-occurrence frequency of terms with surrounding words (such as "shoulder joint dysfunction" often co-occurring with "rehabilitation treatment," while "loss of shoulder joint mobility" often co-occurring with "surgical indication") to help determine the semantic focus.

[0040] In this embodiment, the technical problem to be solved is work-related injury analysis, which needs to be distinguished from other non-work-related injuries. Therefore, it is necessary to identify and map work-related injury characteristics. The specific steps are as follows: 1. Extraction of work-related injury-specific features Construct a corpus of work-related injury information and label key features (such as "working hours", "workplace", "occupational disease catalog").

[0041] The BiLSTM-CRF classification model is used to perform sequence labeling on the input text to identify entities related to work-related injuries.

[0042] Based on the criteria for work-related injury determination in the "Regulations on Work-Related Injury Insurance", regular expression matching rules were designed.

[0043] 2. Multimodal data fusion By combining multi-source data such as medical records (e.g., diagnostic reports), work injury determination application forms, and surveillance videos, work injury-related features are extracted through a cross-modal attention mechanism.

[0044] S2.1.3: Store the knowledge association network in the form of a graph database.

[0045] In one specific implementation, a graph database (such as Neo4j) is constructed in the knowledge storage and query module to store entities and relationships, supporting complex queries (such as "query all injury types that cause knee joint dysfunction and their corresponding disability levels"). Knowledge is stored in the graph database in a graph structure to form a work injury knowledge graph. In the graph structure, structured information related to work injury conditions is used as nodes, and the relationships between structured information are used as edges.

[0046] The entity includes the type of injury, anatomical location, functional impairment, disability level, assessment methods, and assessment tools. Specifically, it includes the following: Types of injury: such as fractures, nerve damage, joint dislocations, burns, etc.

[0047] Anatomical locations: such as the shoulder joint, knee joint, wrist joint, spine, etc.

[0048] Functional impairments: such as motor dysfunction, sensory dysfunction, and limited ability to perform daily living activities.

[0049] Disability level: Level 1 to Level 10 (based on standard grading).

[0050] Assessment methods: joint range of motion measurement, muscle strength scoring, imaging examinations (X-ray / CT / MRI), and scale assessments (such as the Barthel Index).

[0051] Assessment tools: such as goniometers, muscle strength testers, imaging equipment, etc.

[0052] The relationship includes the following: Caused by: Injury type → functional impairment (e.g., "fracture → resulting in → knee joint movement dysfunction").

[0053] Impact: Functional impairment → Disability level (e.g., "Knee joint movement dysfunction → Impact → Disability level eight").

[0054] Usage: Assessment method → ​​Assessment tool (e.g., "Joint range of motion measurement → Usage → Goniometer").

[0055] Includes: Disability level → Grading details (e.g., "Level 8 Disability → Includes → Loss of function of one limb ≥ 25%").

[0056] The attributes include the following: Numerical attributes: Range of Motion (ROM), muscle strength grade, and percentage of functional loss. ROM is a quantitative indicator of a joint's range of motion, referring to the angle of rotation of a joint from its initial position (usually anatomically neutral) to its maximum extension or flexion position, measured in degrees. It is an important basis for assessing joint function, diagnosing musculoskeletal injuries or diseases, and developing rehabilitation plans. Text-based attributes: definition of disability level, compensation standard clauses.

[0057] S2.2: Embed the constructed work injury knowledge graph into the work injury analysis model.

[0058] In one specific implementation, the constructed work injury knowledge graph is embedded into the work injury analysis model, and when invoked, a graph query language (such as Cypher) is used to retrieve related data in real time.

[0059] S3: Using a work injury analysis model with a work injury knowledge graph, dynamic injury reasoning is performed based on the path matching between the injury information to be assessed and the work injury knowledge graph, to obtain the disability level assessment result and an interpretable reasoning path.

[0060] S3.1: Preprocess the injury information to be assessed to obtain standardized injury data.

[0061] In this embodiment, the injury input module is used to convert the input injury information (structured / unstructured) to be identified into standardized data that can be understood by a knowledge graph.

[0062] S3.1.1: Analyze the injury information to be identified.

[0063] S3.1.2: Map the injury information to be identified to standard terminology based on the analysis results.

[0064] Structured data is extracted directly. The time format is standardized (e.g., YYYY-MM-DD HH:MM) and the part terminology is standardized (e.g., "right hand" is mapped to "upper limb-right hand") through the rule engine.

[0065] Unstructured text is parsed. Jieba segmentation and BERT-CRF model are used to identify entities such as time, location, injury location, and action. Dependency parsing is used to analyze the relationships between entities, such as "operating the punch press → causing → right hand to be crushed", forming an "event-injury" causal chain, which is stored as "causal relationship edges" in the graph.

[0066] S3.2: Perform work injury knowledge graph path matching based on standardized injury data to obtain inference paths and inference results.

[0067] In this embodiment, the injury reasoning module utilizes a work injury knowledge graph to achieve intelligent matching and dynamic reasoning between injury characteristics and relevant clauses. When performing work injury knowledge graph path matching based on standardized injury data, if multiple paths match the injury, the path superposition effect is determined according to priority settings and anatomical weights.

[0068] Based on standardized injury data, the system searches for perfectly matching clause nodes in the graph. If the injury is completely consistent with the description of a clause, the system returns the content of that clause and the disability level.

[0069] If the injury matches multiple clauses, analyze the priority of the clauses (e.g., "complete loss of function" takes precedence over "partial loss"), combine the anatomical weights (e.g., the proportion of upper limb function) and the cumulative effect of the rules, and determine the final disability level.

[0070] Returns the content of the disability clause and the disability level, and generates an interpretable reasoning path.

[0071] S3.3: Detect whether the injury information to be identified meets the special rules, and make inferences and corrections based on the detection results of the special rules.

[0072] The specific steps are as follows: First, establish priority rules, including the principle of priority based on rank and the principle of priority based on special clauses.

[0073] The priority principle is that when an injury matches multiple clauses, the clause with the higher disability level takes precedence (e.g., level nine takes precedence over level ten). If multiple clauses have the same level, the promotion principle is triggered: two or more clauses with the same level can be promoted one level (e.g., two level nine clauses can be promoted to level eight).

[0074] The priority principle for special clauses is that occupational diseases (such as pneumoconiosis) take precedence over ordinary mechanical injuries; clauses on complete loss of function take precedence over clauses on partial loss of function (such as "complete loss of knee joint function" taking precedence over "partial limitation of movement").

[0075] Then, the anatomical weights are combined according to the principle of priority of grade and the principle of priority of special clauses.

[0076] 1. Weighting of Functional Aspects Based on the "Standards for the Assessment of Work Capacity" and medical anatomy, the weights for each body part are set as follows: In this embodiment, the upper limbs account for 60% of the overall function, and the lower limbs account for 40%. Upper limb breakdown: shoulder joint 45%, elbow joint 65%, wrist joint 18% (weighted by function percentage).

[0077] Lower limbs are further divided into: hip joint 30%, knee joint 50%, and ankle joint 20%.

[0078] 2. Calculation of multiple site damage superposition The total loss of function ratio was calculated using the weighted average method: If each upper limb is injured by 50%, then the total loss percentage = (50% × 60%) + (50% × 60%) = 60% → corresponding to a level 6 disability.

[0079] If different body parts are involved (such as upper limbs + lower limbs), the weights of each body part are calculated separately and then summed.

[0080] Finally, the rule overlay effect is generated. The rule overlay logic includes: Single item priority: First, assess the level of each clause independently, and take the highest level as the base level.

[0081] Promotion adjustment: If multiple clauses are at the same level, they will be promoted by one level (e.g., two clauses at level nine → level eight).

[0082] Functional superposition: The total functional loss ratio is calculated by combining anatomical weights and matched with the grade range in the "Standards for the Assessment of Work Capacity" (e.g., total loss ≥ 50% corresponds to grade seven).

[0083] Finally, this embodiment uses the analysis result output module to output the clause content, disability level, and interpretable reasoning path.

[0084] Example 2: Embodiment 2 of the present invention provides a knowledge graph-based auxiliary analysis system for work-related injuries, comprising: The data acquisition module is configured to acquire structured information related to work-related injuries and information on injuries to be assessed. The knowledge graph construction module is configured to build a work injury knowledge graph based on structured information related to work injury conditions, and embed the constructed work injury knowledge graph into the work injury condition analysis model. The injury reasoning module is configured to use a work injury analysis model with a work injury knowledge graph to perform dynamic injury reasoning based on the path matching between the injury information to be identified and the work injury knowledge graph, and obtain the disability level assessment result and an interpretable reasoning path.

[0085] Example 3: Embodiment 3 of the present invention provides a computer-readable storage medium storing a computer program adapted for loading by a processor and executing the steps of the knowledge graph-based auxiliary analysis method for work-related injuries as described in Embodiment 1 of the present invention.

[0086] Example 4: Embodiment 4 of the present invention provides a computer device, the device comprising: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the steps in the knowledge graph-based auxiliary analysis method for work-related injuries as described in Embodiment 1 of the present invention.

[0087] The steps and methods involved in Examples 2, 3 and 4 above correspond to those in Example 1. For specific implementation details, please refer to the relevant description section of Example 1.

[0088] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application. In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium that a computer can access or a data processing device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, an optical medium, or a semiconductor medium, etc. The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A knowledge graph-based auxiliary analysis method for work injury conditions, characterized in that, The method comprises the following steps: obtaining structured information related to work injury and information to be identified; constructing a work injury knowledge graph according to the structured information related to work injury, and embedding the constructed work injury knowledge graph into a work injury condition analysis model; performing dynamic injury reasoning according to the path matching condition between the information to be identified and the work injury knowledge graph by using the work injury condition analysis model with the work injury knowledge graph, to obtain an injury grade identification result and an interpretable reasoning path. 2.The knowledge graph-based work injury condition auxiliary analysis method of claim 1, wherein, The structured information related to work injury includes clause numbers of relevant standards, applicable injury clauses, injury grade settings, and correlation rules. 3.The knowledge graph-based work injury condition auxiliary analysis method of claim 1, wherein, The specific steps of constructing the work injury knowledge graph according to the structured information related to work injury are as follows: performing knowledge extraction on the structured information related to work injury; fusing knowledge according to the relationship between the knowledge, and constructing a knowledge correlation network; storing the knowledge correlation network in the form of a graph structure into a graph database. 4.The knowledge graph-based work injury condition auxiliary analysis method of claim 3, wherein, In the graph structure, the structured information related to work injury is taken as a node, and the correlation relationship between the structured information is taken as an edge. 5.The knowledge graph-based work injury condition auxiliary analysis method of claim 1, wherein, The specific steps of performing dynamic injury reasoning according to the path matching condition between the information to be identified and the work injury knowledge graph by using the work injury condition analysis model with the work injury knowledge graph are as follows: performing preprocessing on the information to be identified, to obtain standardized injury data; performing path matching of the work injury knowledge graph based on the standardized injury data, to obtain a reasoning path and a reasoning result; detecting whether the information to be identified meets special rules, and performing reasoning correction according to the detection condition of the special rules. 6.The knowledge graph-based work injury condition auxiliary analysis method of claim 5, wherein, The preprocessing step on the information to be identified comprises: analyzing the information to be identified; mapping the information to be identified to standard terms according to the analysis result.

7. The knowledge graph-based work injury condition auxiliary analysis method according to claim 6, characterized in that, When multiple paths are matched in the path matching of the work injury knowledge graph based on the standardized injury data, the path superposition effect is determined according to the priority setting and the anatomical weight.

8. An injury condition auxiliary analysis system based on a knowledge graph, characterized in that, The method comprises the following steps: a data acquisition module configured to obtain structured information related to work injury and information to be identified; a knowledge graph construction module configured to construct a work injury knowledge graph according to the structured information related to work injury, and embed the constructed work injury knowledge graph into a work injury condition analysis model; an injury reasoning module configured to perform dynamic injury reasoning according to the path matching condition between the information to be identified and the work injury knowledge graph by using the work injury condition analysis model with the work injury knowledge graph, to obtain an injury grade identification result and an interpretable reasoning path.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program is suitable for being loaded and executed by the processor and implements the knowledge graph-based work injury condition auxiliary analysis method in any one of claims 1-7.

10. A computer device, comprising: The method comprises the following steps: a processor adapted to execute a computer program; a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program is executed by the processor to implement the knowledge graph-based work injury condition auxiliary analysis method in any one of claims 1-7.

Images