Method and system for constructing aviation equipment environment test knowledge graph based on semantic consistency

Abstract

This invention provides a method and system for constructing a knowledge graph for aviation equipment environmental testing based on semantic consistency. It relates to the field of aviation equipment testing environment establishment and testing technology. The method includes: S1, acquiring standard texts for aviation equipment environmental testing and constructing a standard knowledge system base; S2, extracting entities, entity relationships, and entity attributes of the aviation equipment environmental testing standards based on a large language model; S3, merging semantically identical aviation equipment environmental testing standard entities based on semantic vector similarity detection; S4, achieving consistency detection of aviation equipment environmental testing standard entities based on graph relationships and attribute similarity; S5, fusing the results of steps S3 and S4 to determine the aviation equipment environmental testing standard entities and constructing a knowledge graph for aviation equipment environmental testing standards. This invention constructs the extraction results of entities, relationships, and attributes of aviation equipment environmental testing standards using a large language model; and generates a high-quality, highly consistent knowledge graph for aviation equipment environmental testing standards based on semantic consistency enhancement.

Description

Technical Field

[0001] This invention relates to the field of aviation equipment test environment establishment and testing technology, specifically to a method and system for constructing an aviation equipment environment test knowledge graph based on semantic consistency. Background Technology

[0002] Knowledge graphs, as a core technology for structured knowledge representation, have demonstrated immense value in fields such as intelligent search, decision support, and semantic understanding. They effectively organize and reason about complex knowledge by transforming real-world concepts, entities, and their relationships into graph structures of triples (entity-relationship-entity or entity-attribute-value). In the field of environmental testing standards for aviation equipment, the construction of knowledge graphs can significantly improve the accessibility and application efficiency of standard information, such as supporting rapid retrieval of standard clauses, compliance checks of technical indicators, and cross-domain knowledge correlation analysis. However, environmental testing standards for aviation equipment are typically published in PDF format, and this unstructured or semi-structured document format poses a significant challenge to automated processing. Existing methods largely rely on Optical Character Recognition (OCR) technology to directly extract text content. However, the OCR process is susceptible to document layout, formula symbols, or table structures, introducing problems such as recognition errors, formatting chaos, and even semantic distortion, leading to a significant decrease in the accuracy of subsequent knowledge extraction. In addition, the environmental testing standards for aviation equipment contain a large number of technical terms, normative expressions and cross-chapter citations. Traditional natural language processing (NLP) tools are unable to effectively analyze their deep semantics, and there is an urgent need for a preprocessing mechanism that can take into account both text quality repair and content structuring.

[0003] Currently, knowledge graph construction methods face several bottlenecks in vertical domain applications: First, rule-based or traditional machine learning-based information extraction methods heavily rely on manually labeled data and domain feature engineering. However, in professional scenarios such as aviation equipment environmental testing standards, high-quality training data is scarce and labeling costs are high, resulting in insufficient model generalization ability. Second, existing entity, relation, and attribute extraction methods for aviation equipment environmental testing projects, equipment, and parameters often use independent modules and lack a unified framework, which can easily lead to inconsistent extraction results, entity redundancy, or relation omissions. For example, entity recognition for aviation equipment environmental testing projects, equipment, and parameters may ignore synonym variants, relation extraction may struggle to capture implicit semantic associations, and attribute extraction may fail to accurately handle complex information in long text descriptions. While Large Language Models (LLMs) offer a novel approach to zero-shot or few-shot information extraction due to their powerful contextual understanding and generation capabilities, their direct application to knowledge graph construction for aviation equipment environmental testing standards still has significant limitations. Firstly, LLMs are highly sensitive to the quality of input text, and OCR errors can accumulate and amplify model bias. Secondly, aviation equipment environmental testing standards involve highly structured knowledge systems, such as classification systems, entity types, and relational settings, requiring targeted design of prompts and knowledge guidance strategies. Existing methods lack dynamic construction and adaptation mechanisms for these domain knowledge systems. Furthermore, the post-processing verification and fusion of extracted results are often neglected, leading to issues such as entity misalignment, relational breaks, or attribute conflicts in the knowledge graph. Therefore, developing a method that integrates text preprocessing, knowledge system construction, LLM-driven extraction, and result optimization for extracting and constructing knowledge graphs for aviation equipment environmental testing standards is a key requirement for overcoming existing technological bottlenecks. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide a method for constructing a knowledge graph for aviation equipment environmental testing based on semantic consistency. This method dynamically develops a knowledge system library for aviation equipment environmental testing standards across different fields, including entity types, relationships, and attribute settings. By using a large language model to construct entity extraction, entity relationship extraction, and entity attribute extraction methods for aviation equipment environmental testing standards, the extraction results of entities, relationships, and attributes of aviation equipment environmental testing standards are obtained. Finally, a high-quality, highly consistent knowledge graph for aviation equipment environmental testing standards is generated based on semantic consistency enhancement verification.

[0005] Specifically, on the one hand, the present invention provides a method for constructing a knowledge graph for aviation equipment environmental testing based on semantic consistency, which includes the following steps: S1: Optical character recognition of aviation equipment environmental testing standard documents yields the input text of aviation equipment environmental testing standards; pre-setting the entity types, entity relationships, and entity attributes of aviation equipment environmental testing standards; aviation equipment environmental testing standard entities include: test items, test equipment, and test parameters; constructing a knowledge system base for aviation equipment environmental testing standards; S2: Construct a standard entity extraction method for aviation equipment environmental testing using a large language model, extract the standard entities, entity relationships, and entity attributes of aviation equipment environmental testing; supplement the standard entity extraction results for aviation equipment environmental testing based on the entity relationship and entity attribute extraction results; S3: Perform semantic vectorization of standard entities for environmental testing of aviation equipment, calculate semantic similarity of standard entities for environmental testing of aviation equipment, and classify and filter entities; use semantic vector similarity to detect standard entity pairs for environmental testing of aviation equipment, and merge standard entities for environmental testing of aviation equipment with the same semantics. S4: Based on graph relationships and attribute similarity, achieve consistency detection of aviation equipment environmental test standard entities; utilize the relational topology and attribute characteristics of aviation equipment environmental test standard entities in the knowledge graph to determine the association relationships of test equipment used in test projects, test parameters of test projects, and test parameters containing physical attribute characteristics; detect and merge aviation equipment environmental test standard entities with similar relationships to make the knowledge graph structure consistent. S5: Calculate the cluster similarity of entities in the aviation equipment environmental testing standard; fuse the semantic vector similarity results of the aviation equipment environmental testing standard entities in step S3 with the entity graph relationship similarity results of the aviation equipment environmental testing standard in step S4; determine the final list of aviation equipment environmental testing standard entities and obtain the aviation equipment environmental testing standard knowledge graph.

[0006] On the other hand, the present invention provides a knowledge graph construction system based on a semantic consistency-based method for constructing a knowledge graph for environmental testing of aviation equipment, which includes: an entity recognition module, a relationship recognition module, an attribute recognition module, and a data processing module; The entity recognition module extracts the prompt word template of the aviation equipment environmental test standard entity, uses a large language model to extract the aviation equipment environmental test standard entity from the input text, extracts all aviation equipment environmental test standard entities that meet the conditions, generates a list of aviation equipment environmental test standard entities, and realizes the automatic recognition of key entities in the aviation equipment environmental test standard text. The relationship recognition module establishes the associations between entities in the aviation equipment environmental testing standard, forming relationship edges in the knowledge graph, enhancing the structured representation of knowledge, and extracting semantic relationships between entities in the aviation equipment environmental testing standard from the aviation equipment environmental testing standard text and the identified aviation equipment environmental testing standard entities. The attribute recognition module includes detailed information on the entities of the aviation equipment environmental testing standards, improves the feature descriptions of the entities of the aviation equipment environmental testing standards in the knowledge graph, thereby supporting more granular knowledge query and analysis, and is responsible for extracting the attribute information of the aviation equipment environmental testing standards entities from the aviation equipment environmental testing standard text and the identified aviation equipment environmental testing standard entities. The data processing module is responsible for integrating, cleaning, and semantically optimizing the entity, relation, and attribute data of the aviation equipment environmental testing standards output by the aforementioned modules.

[0007] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention is based on a large language model to realize the knowledge graph construction method of aviation equipment environmental test standards. In view of the problem that the knowledge system in the field of aviation equipment environmental test standards lacks dynamic adaptation, the present invention dynamically constructs a knowledge system library of entity types, relationships and attribute settings of aviation equipment environmental test items, test equipment and test parameters through a large language model, thereby realizing targeted guidance for aviation equipment testing, improving the accuracy and generalization ability of knowledge extraction, and reducing the cost of manual annotation and the dependence on feature engineering.

[0008] (2) This invention addresses the inconsistency caused by the independent extraction modules of entities, relationships and attributes of environmental test items, test equipment and test parameters of aviation equipment. It performs collaborative extraction based on a unified large language model framework and achieves high-precision information extraction through customized prompt word templates. This ensures the semantic consistency and integrity between entities, relationships and attributes of environmental test items, test equipment and test parameters of aviation equipment, and reduces data redundancy and omissions.

[0009] (3) This invention addresses the redundancy and alignment issues caused by synonymous entities. Through traversal matching and semantic disambiguation of the large language model in the post-processing step, it verifies and merges entities that do not match aviation equipment environmental test items, test equipment and test parameters, effectively improving the alignment quality and overall consistency of the knowledge graph and supporting more efficient knowledge reasoning and application. Attached Figure Description

[0010] Figure 1 The accompanying figure is an abstract of the present invention regarding a method for constructing an environmental testing knowledge graph for aviation equipment based on semantic consistency. Figure 2 A flowchart illustrating a method for constructing an aviation equipment environmental testing knowledge graph based on semantic consistency, as described in this invention. Figure 3 This is the ontology structure diagram of the common knowledge system built based on the environmental testing standards for aviation equipment in this invention; Figure 4 To construct the entity relationship connection diagram for the environmental testing part of this invention; Figure 5 This is a diagram of the entity extraction mechanism driven by the large language model of the present invention; Figure 6 This is a comparison chart of the experimental results of entity extraction and relation extraction of this invention with GraphRAG and LightRAG. Detailed Implementation

[0011] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.

[0012] This invention proposes a method for constructing a knowledge graph for environmental testing of aviation equipment based on semantic consistency, such as... Figure 1 As shown, the process involves: acquiring the text of the aviation equipment environmental testing standard and constructing a standard knowledge system base; extracting entities, entity relationships, and entity attributes of the aviation equipment environmental testing standard based on a large language model; merging semantically identical aviation equipment environmental testing standard entities based on semantic vector similarity detection; performing consistency detection of aviation equipment environmental testing standard entities based on graph relationships and attribute similarity; and fusing and determining the aviation equipment environmental testing standard entities to construct an aviation equipment environmental testing standard knowledge graph. The process includes the following steps: Step S1: Obtain the environmental testing standard documents for aviation equipment and preprocess them; construct a knowledge system base for environmental testing standards for aviation equipment.

[0013] Step S11: Obtain the aviation equipment environmental testing standard document in portable PDF format, perform optical character recognition (OCR) to obtain the original recognized aviation equipment environmental testing standard text. Using preset prompts to guide the large language model in processing the environmental testing standard text for aviation equipment. Error correction and format standardization were performed to obtain the preprocessed input text for the aviation equipment environmental testing standard. The PDF format aviation equipment environmental testing standard document in this embodiment of the invention contains tables, line graphs, flowcharts, and other types of charts; for these charts, a chart information extraction step is performed to convert them into text that is easy to directly input into a large language model; specifically, the following steps are included: Step S111: Segment the aviation equipment environmental test standard document to obtain standard image blocks; This embodiment of the invention uses the PyMuPDF library function to convert the pages of the aviation equipment environmental test standard document into images, and uses optical character recognition (OCR) technology for recognition. This method uses the PaddleOCR model to obtain all standard image blocks of the aviation equipment environmental test standard document, including bounding boxes and text content, and records each text region as a standard text block; the parts of the entire aviation equipment environmental test standard document that do not belong to any text region are extracted and filtered, and used as standard image blocks. The PyMuPDF library function in this embodiment of the invention is a Python library specifically designed for processing PDF, XPS, and other documents. Its core function is to convert PDF pages into raster images. The PaddleOCR model is an open-source, multilingual optical character recognition (OCR) tool library that can accurately recognize text in images. Inputting an image, it can output text content, location, and confidence level, supporting multilingual, table, and layout analysis, combining high accuracy and high performance.

[0014] Step S112: Classify the standard image blocks in the aviation equipment environmental test standard documents; based on the standard image blocks obtained in step S111, classify them; in the aviation equipment environmental test standard documents, the commonly appearing chart types are tables, flowcharts, and line graphs. The standard image blocks are divided into the following four categories: tables, flowcharts, line graphs, and others. In this embodiment, a lightweight standard image classification model was trained based on the deep convolutional neural network model ResNet50D. In testing, the classification accuracy was no less than 95%. Based on the classification results, different processing was applied to different types of table-type, flowchart-type, and line graph-type standard image blocks. The core function of the deep convolutional neural network model ResNet-50D is to perform high-precision image feature extraction and classification; through its optimized residual module structure, this model can effectively capture complex features in images. As a high-performance backbone network in computer vision tasks such as image recognition, object detection, and semantic segmentation, it demonstrates excellent representation capabilities and generalization performance.

[0015] Step S113: For table-type standard image blocks in the aviation equipment environmental testing standard document, convert them into grayscale images and binarize them. Then, use the PaddleOCR model to recognize the content of the table-type standard image blocks to obtain the text, classify them by line, and then classify the aviation equipment environmental testing standard text. The input is fed into the large language model.

[0016] Step S114: Extract flowchart-type standard image block information from the aviation equipment environmental testing standard document, specifically including the following steps: Step S1141: Preprocessing and Contour Recognition of Flowchart-type Standard Image Blocks in the Aircraft Equipment Environmental Testing Standard Document; convert them into grayscale images and binarize them; perform contour recognition based on the Suzuki-Abe contour tracking algorithm to obtain the contour set of the flowchart-type standard image blocks. The Suzuki-Abe contour tracking algorithm is used to extract the contour boundaries of target objects from binary images. By taking a binary image as input, it outputs a set of hierarchically organized external and internal contour sequences, which can accurately describe the topological relationship of the object's shape.

[0017] Step S1142: Node detection of flowchart-type standard image blocks in the aviation equipment environmental testing standard document; calculation of the bounding box of each flowchart-type standard image block. The Douglas-Peucker algorithm is used to approximate the outline polygon, obtaining an approximate number of vertices for standard flowchart image blocks. If the approximate number of vertices If the angle is close to 90°, it is determined to be a rectangle. Rectangular nodes represent processes or operations, used to indicate a specific execution step or task. If the fit is good, such as an ellipse or circle, it is determined to be the start or end node of an ellipse, used to indicate the beginning and end of the process. If it is a rhombus with four sides but a rotation angle close to 45°, it is determined to be a decision node. This decision node usually contains a yes / no question and guides the process to different branches based on the result. Combining these two, the output of this step is: a candidate set of nodes for standard flowchart image blocks. ; For the first A flowchart-type standard image block node boundary; This refers to the standard image block shape for flowcharts. The Douglas-Peucker algorithm is a geometric simplification algorithm for vector curve data compression; it takes a curve consisting of ordered vertices as input and outputs a simplified curve consisting of key vertices that retains the original geometry to the maximum extent within a specified tolerance.

[0018] Step S1143: Perform edge detection and orientation determination for flowchart-type standard image blocks in the aviation equipment environmental testing standard document; based on the contour set of the binarized flowchart-type standard image blocks in step S1141 and the node candidate set of the flowchart-type standard image blocks in S1142. Based on the Zhang-Suen algorithm, the thin line structure is extracted, and the connecting area is compressed into a standard skeleton map with a width of one pixel. The connected structure is preserved. Then, based on the pixel neighborhood degree, the endpoint degree of the standard skeleton graph is identified as 1 and the intersection degree is ≥3. Finally, the pixel path set of each connection is extracted using the depth-first search (DFS) algorithm. Among them, the Zhang-Suen thinning algorithm is a parallel thinning algorithm for binary images. This algorithm takes a binary image as input and outputs its skeleton representation with a single pixel width, effectively preserving the original shape's topological structure and connectivity. Depth-First Search (DFS) is an algorithm for traversing or searching tree and graph structures. Starting from a specified initial node, this algorithm systematically explores every branch until the end, outputting a depth-first traversal sequence of nodes or a specific target node found through the search. The Douglas-Peucker algorithm is used near both ends of each path to identify arrow shapes, and the direction of the connection is determined by the direction vector of the triangle's tip. The endpoints of the line segments in the standard flowchart image block are... Mapped to the boundary of adjacent flowchart class standard image block nodes If the line segment end of a flowchart-type standard image block is adjacent to the node boundary of a flowchart-type standard image block... distance If the value is connected to that node, then the shortest Euclidean distance to the standard image block of the flowchart class is obtained. for: ; in, The shortest Euclidean distance for standard image blocks in flowcharts; This is a standard image block for flowcharts; For set Standard node elements in; To find the minimum value of the function; For the first A flowchart-type standard image block node boundary; Number the standard image blocks for flowcharts; This is a set of standard image block node boundaries for flowchart classes.

[0019] Based on the shortest Euclidean distance of standard image blocks in flowcharts This process matches path endpoints to corresponding nodes, establishing a semantic association between lines and nodes. If an arrow exists within a line, its direction is determined by the arrow's pointing direction; if the arrow is located at one end of the line and points to another node, a directed edge is established. Record the initial edge set of standard image blocks for flowcharts. Includes geometric path and arrow direction / confidence level.

[0020] Step S1144: Determine the text and node types of the flowcharts in the aviation equipment environmental testing standard documents; for the node candidate set of the flowchart-type standard image blocks obtained in step S1142. , including: the A flowchart-type standard image block node boundary Standard image block shape for flowcharts Text areas are cropped from each node of a standard flowchart-type image block. The input text of the PaddleOCR model was fed into the PaddleOCR model and the recognition result was: the input text of the aviation equipment environmental test standard. With confidence level , in the form of The final result is the node set of text confidence scores for standard flowchart image blocks. ,in, This indicates the geometric position transformation information of the text region of the flowchart-type standard image block node in the image; if the flowchart-type standard image block shape obtained in step S1142... If the ellipse or text contains "start / end", then That is, the start or end node; if the flowchart is a standard image block shape If the diamond or text contains the question / interrogative word "Is / Is / OK?", then That is, a judgment or decision node; otherwise, it is considered as... This refers to process or operation nodes; this step yields a set of nodes for standard flowchart image blocks. .

[0021] Step S1145: Reconstruct the graph structure of the flowchart-type standard image blocks in the aviation equipment environmental testing standard document; based on the node set of the flowchart-type standard image blocks obtained in step S1144. The preliminary edge set of the flowchart-class standard image blocks obtained in step S1143 Map the path of each geometric edge to the edge on the flowchart. If the path endpoint maps to the text confidence node of a standard image block in a flowchart class. and Then establish edges; if the path contains multiple inflection points, retain the sequence of points along the polyline path. For visualization purposes; to determine node-to-node connection judgments or decision branches, we attempt to extract near-text labels from standard flowchart image blocks on the edges: perform small-scale text detection near the connections to find edge labels such as "yes / no", the result of which is the near-text label of standard flowchart image blocks. ; Construct the edge set of standard image blocks for flowcharts Each edge has a label and a geometric path; each edge of a standard image block in a flowchart class. It is usually represented as a quadruple as: ; in, For the first The edges of a standard image block of a flowchart class constitute the edge set of the standard image block of the flowchart class. ; The text confidence node ID for the standard image block of the flowchart at the starting point; The text confidence node ID for the standard image block of the flowchart at the endpoint is determined by node matching. The geometric path of the connecting line is a sequence of pixel coordinates. The OCR results are near the text labels of standard image blocks in the flowchart category, such as "Yes / No".

[0022] Step S1146: Convert the graph structure of the flowchart-type standard image blocks in the aviation equipment environmental testing standard document into natural language; based on the node set and edge set of the flowchart-type standard image blocks obtained in step S1145. Construct a directed graph of standard image blocks for flowcharts. Records the node collection of standard image blocks for flowcharts in JSON format. Sum of edges Combined with a preset template, the input large language model will generate a directed graph. Paraphrase into a text paragraph.

[0023] Step S115: Extract standard image block information of line graph type from the aviation equipment environmental test standard file. Binarization is required first, followed by applying morphological opening operations to remove noise points and connect broken lines. Morphological opening is a nonlinear filtering operation used in image processing; by inputting a binary image and a structuring element, it outputs a processed image, effectively eliminating minor noise, separating adhered objects, and smoothing target contours while maintaining the main geometric shape.

[0024] Step S1151: Detect the coordinate axes of the standard image blocks of line graphs in the aviation equipment environmental testing standard document; use the probabilistic Hough LinesP transform to detect straight line segments in the standard image blocks of line graphs, and perform DBSCAN clustering on the detected line segments according to their angles to identify approximately horizontal and vertical line segment groups. The longest horizontal line segment is used as the x-axis, and the longest vertical line segment is used as the y-axis. The probabilistic Hough LinesP transform is used to detect straight line segments from images; by inputting a binary image, the output is a set consisting of the coordinates of the start and end points of the line segments, which can effectively locate the linear geometric structures existing in the image. DBSCAN is a density-based spatial clustering algorithm; by inputting a set of data points and neighborhood parameters, it can discover clusters of arbitrary shapes and identify noise points, and its output is the cluster label or noise marker to which each data point belongs.

[0025] Step S1152: Identify the scale labels of standard image blocks in the line graph type in the aviation equipment environmental testing standard document; perform text recognition using the PaddleOCR model; first, determine the label area based on the coordinate axis position, and then use projection analysis to locate individual character areas; configure and optimize the parameters of the PaddleOCR model, setting it to only recognize numbers and common symbols to improve recognition accuracy. Projection analysis is a method for text layout analysis in optical character recognition (OCR); by statistically analyzing the pixel projection distribution of the image in the horizontal and vertical directions, it effectively locates text lines, segments characters, and identifies the layout structure, laying the foundation for subsequent recognition.

[0026] Step S1153: Extract data points from standard image blocks of polyline graphs in the aviation equipment environmental testing standard document; extract the polyline contour using the Suzuki-Abe contour tracking algorithm, and then apply the Zhang-Suen skeleton thinning algorithm to obtain polylines with a single pixel width. Sample points along the thinned polylines at fixed intervals, and determine the inflection points of the polylines through nearest neighbor search.

[0027] Step S1154: Perform coordinate mapping of standard image blocks of the line graph type in the aviation equipment environmental test standard document; establish the mapping relationship between the image pixel coordinate system and the actual data value; identify the scale label value and its position in the image, and use the linear interpolation formula to convert the pixel coordinates of the data point into the actual value as: actual value = minimum value + (pixel coordinate - minimum pixel value) × (maximum value - minimum value) / (maximum pixel value - minimum pixel value).

[0028] Step S1155: Structure the data of the standard image blocks of line graphs in the aviation equipment environmental test standard file; organize the extracted data into JSON format, design a structured prompt word template, embed the extracted data into it, and use a large language model to describe the data as natural language.

[0029] Step S116: In the aviation equipment environmental testing standard document, for other types of images not covered in the aforementioned process, they are directly input into the large language model. The large language model performs information parsing and content summarization, and generates a generalized natural language summary. Example: In this step, the input aviation equipment environmental testing standard text... In China, due to factors such as scanning quality, page layout, and character similarity, OCR recognition results often contain various errors, such as confusion between the number "0" and the letter "O", incorrect paragraph heading positions, redundant page numbers, and inconsistencies in the hyphen symbol in standard numbering. This is achieved by analyzing the text of the aviation equipment environmental testing standard. Input into the large language model for error correction, correcting character recognition errors, adjusting paragraph structures, deleting meaningless fragmented texts, and unifying the symbols in the standard numbers. For example, for the common connector "-" in standard documents, it is often recognized as "one" or "—" during recognition. Through the processing of the large language model, it is uniformly standardized as a short horizontal line "-" to ensure the unity and uniqueness of the entity recognition results. By combining rule guidance and semantic understanding, while maximizing the retention of the original text information, the structural standardization and character accuracy of the text are significantly improved.

[0030] Step S12: Preset a list of entity types for aviation equipment environmental test standards; the entity types include but are not limited to test items, test equipment, test parameters, test methods, standards, articles, and industry fields; using the preprocessed aviation equipment environmental test standard text as input, classify the aviation equipment environmental test standard text through step S11 to extract aviation equipment environmental test standard entity tags and determine the aviation equipment environmental test standard entity types.

[0031] Step S121: Classify the content of the aviation equipment environmental test standards; based on the preprocessed aviation equipment environmental test standard input text completed in step S11 , classify the aviation equipment environmental test standards according to the main content of the aviation equipment environmental test standards to obtain the aviation equipment environmental test standard text classified based on content ; in the subsequent process of constructing the knowledge system, the aviation equipment environmental test standard knowledge system constructed by this method is usually divided into two parts. One part is the common knowledge system of aviation equipment environmental test standards, including information such as chapter structures and issuing entities, as Figure 3 shown as the ontology of the common knowledge system constructed based on aviation equipment environmental test standards; the other part is the knowledge system of the aviation equipment environmental field targeting the content.

[0032] Step S122: Extract aviation equipment environmental test standard entity tags; according to the aviation equipment environmental test standards classified in step S121, select several texts in the same type of standards, input them into the large language model, and combine the preset prompt words to construct the aviation equipment environmental test standard entity tag and entity category table under this content classification as shown in Table 1.

[0033] Table 1 Partial Entity Tag Table Salt spray test chamber Types of test equipment Vibration table Types of test equipment temperature Test parameter types humidity Test parameter types High temperature test Test Project Types Low temperature test Test Project Types Step S123: Determine the entity types of the aviation equipment environmental testing standards; based on the entity tags of the aviation equipment environmental testing standards obtained in step S122, and combined with the input text of the aviation equipment environmental testing standards in step S11 and the preset prompt words input large language model, obtain the list of entity types in the aviation equipment environmental testing standards. List of entity types in the embodiments The format is as follows: The main body is the name of the subject appearing in the text, usually the drafting body, etc.; Test parameters: the physical quantities that need to be controlled or measured in the test; Test sample: the product or component being tested.

[0034] Step S13: Pre-set the entity relationships of the aviation equipment environmental testing standards; the types of entity relationships include, but are not limited to, including, containing, adopting, having characteristics, composed of, applicable to, and published; based on the list of aviation equipment environmental testing standard entity types obtained in step S12, and in conjunction with the entity type list, input the aviation equipment environmental testing standard text. Construct appropriate prompt words, input them into a large language model, and obtain the entity relationships of aviation equipment environmental testing standards. This relationship setting should include a relationship description, the entity setting domain of the relationship (i.e., the subject entity of the relationship), and the entity value domain of the relationship (i.e., the object entity of the relationship).

[0035] Table 2 Relationship Setting Data table Include Describe the hierarchy and structure between standard documents or chapters, or the specific operational steps included in a process or procedure. Standards, chapters, test items, test methods Standards, chapters, entries, test methods It has characteristics Describe the salient features of a test item, sample, or equipment. Test items, test samples, test equipment Items, test parameters Step S14: Pre-set the entity attributes of the aviation equipment environmental testing standard; entity attributes include physical quantities used to quantify and describe test conditions and results, including but not limited to equipment size, severity level, temperature, time, humidity, pressure, solution pH value, etc.; based on the list of aviation equipment environmental testing standard entity types obtained in step S12, construct prompt words, and input the entity type list and aviation equipment environmental testing standard text. Combined, they are input into a large language model to obtain the entity attributes of aviation equipment environmental testing standards. The entity attributes of the aviation equipment environmental testing standard in the embodiments. The format is as follows: Test / Method: representing various test methods, procedures or experiments described in the document; Parameter: representing various technical indicators, conditions or configurations involved in the experiment; Identification attributes: such as name, standard number, code, number, etc.

[0036] Step S15: List of entity types for environmental testing standards for aviation equipment obtained from steps S12-S14 Substantive Relationship of Environmental Testing Standards for Aviation Equipment Physical attributes of environmental testing standards for aviation equipment Construct an entity relationship attribute connection graph and a knowledge system base for aviation equipment environmental testing standards, such as... Figure 4 The diagram shown is a connection diagram of entity relationships and attributes related to environmental testing of aviation equipment. Due to the large number of entity relationships and attributes involved in the actual extraction, Figure 4 Only some key relationships and attributes are shown for reference.

[0037] Step S2: Use a large language model to construct a method for extracting standard entities for aviation equipment environmental testing, and extract the standard entities, relationships, and attributes of aviation equipment environmental testing entities.

[0038] Step S21: Extract standard entities for environmental testing of aviation equipment using a large language model, such as... Figure 5 As shown; first, a prompt word template containing task description, entity type list, examples, and output format is input along with the preprocessed text X of the aviation equipment environmental testing standard into the large language model LLM. The final output is a list of entity types in the aviation equipment environmental testing standard containing entity names and entity types. The system uses a large language model to extract entities from the environmental testing standards for aviation equipment, outputting a list of entity types in the standards, including entity names and types. Entity types are predefined as entity types that describe environmental testing of aviation equipment, including test items, test equipment, test parameters, test methods, standards, entries, and industry sectors.

[0039] Step S211: Construct a prompt word template for extracting entities from the aviation equipment environmental testing standard; the prompt words for extracting the entity part of the aviation equipment environmental testing standard mainly consist of the following parts: a large language model task description, a list of entity types in related fields, and the input text of the aviation equipment environmental testing standard. Extract examples and output formats. The task description clearly states that examples need to be referenced, based on the entity type list in the aviation equipment environmental testing standards in step S12. Extract all entities from the input national standard text and return the results according to the output format. List of entity types in the environmental testing standards for aviation equipment. For step S12, input the text of the aviation equipment environmental testing standard. This is obtained in step S11. The core content of this step is the construction of prompt words, which sets the method for constructing the prompt words extracted from entities as follows: Among them, the list of entity types for environmental testing standards for aviation equipment The list of entity types for aviation equipment environmental testing standards obtained in step S12, with entity type prompts. This indicates a list of entity types for environmental testing standards for aviation equipment. Combined with templates, the specific implementation is as follows: Task: Refer to the "Example" to extract entities from the "Input Text" based on the "Entity Type List" and return the results according to the "Output Format".

[0040] The entity type is: <input variable ET>.

[0041] Examples are: <Example 1>, <Example 2>, etc.

[0042] The output format is as follows: "XXX||YYY", where XXX is the entity name and YYY is the entity type. Do not output content outside the specified format.

[0043] Step S212: Extract the text of the environmental testing standards for aviation equipment using a large language model. The list of entity types in the aviation equipment environmental testing standards; after the prompt words are constructed, the entity type prompt words will be... Input text for aviation equipment environmental testing standards Combined, the input is fed into a large language model; this process is denoted as... And record the output result as Specifically: ; in, List of entity types for environmental testing standards for aviation equipment; For large language model functions; Entity type hint words; List of entity types for environmental testing standards for aviation equipment; Input text for environmental testing standards for aviation equipment.

[0044] When using large language models to extract standard entities for environmental testing of aviation equipment, the following requirements must be met: ; in, Extracting parameters for entity types refers to a set of variables or results that maximize the objective function. To find the set E that maximizes the function P(E|IT,ET) among all possible values ​​in the set E, we need to find the set E that maximizes the function P(E|IT,ET). ; This is a collection of entities that provide environmental testing standards for aviation equipment.

[0045] Input text according to the given aviation equipment environmental testing standards. and the list of pre-defined environmental testing standards for aviation equipment. The highest probability of obtaining the standard entity sequence for environmental testing of aviation equipment ,in, This represents the sequence of standard entities for generating the most probable environmental testing of aviation equipment under known text and type constraints. The likelihood; due to the autoregressive nature of large language models, this joint probability can be decomposed into the product of the generation probabilities of each aviation equipment environmental test standard entity during sequence generation, specifically: ; in, List of entities for environmental testing standards for aviation equipment Likelihood; The entity name for the environmental testing standards for aviation equipment; For the standard entity type of aviation equipment environmental testing; For the physical sequence of environmental testing standards for aviation equipment The number of entities in; To generate the first A list of all aviation equipment environmental testing standard entities that have been generated before this aviation equipment environmental testing standard entity.

[0046] This formula describes the input text for a given aviation equipment environmental testing standard. List of Entity Types for Environmental Testing Standards for Aviation Equipment Under these conditions, generate a complete list of standard entities for environmental testing of aviation equipment. The joint probability. The large language model will select the sequence of standard entities for aviation equipment environmental testing with the highest probability given the content. As a result Output the results.

[0047] Step S213: Analyze the output of the large language model, specifically: ; in, This is a sequence of standard entities for environmental testing of aviation equipment. This is the parsing function for entity recognition, used to ensure the consistency of the output results; This is the i-th aviation equipment environmental testing standard entity.

[0048] Because the generation results of large language models have a certain degree of randomness, there is a certain probability that the recognition results will include prompts such as "The following are entity recognition results based on reference text," while all that is needed is a list of entities in the form of binary tuples for aviation equipment environmental testing standards. Therefore, the recognition results need to be parsed. In the parsing function... In this context, regular expressions are used for matching; the resulting matches are tuples of the form "XXX||YYY", representing the entity of the aviation equipment environmental testing standard. for: ; in, For the first Name of an entity that provides environmental testing standards for aviation equipment; For the first Each type of entity is a standard for environmental testing of aviation equipment.

[0049] Step S22: Construct a method for extracting entity relationships for aviation equipment environmental testing standards based on a large language model, and extract the entity relationships for aviation equipment environmental testing standards. The types of entity relationships include, but are not limited to, containing, adopting, having characteristics, composed of, applicable to, and publishing. When extracting the attributes of test parameters and test items, the focus is on extracting the numerical values ​​and units of their associated physical quantities, mainly including the numerical values ​​and units of dimensions, temperature, humidity, pressure, and duration. The large language model is used to extract the entity relationships for aviation equipment environmental testing standards; the relationship types are predefined as including, adopting, having characteristics, composed of, applicable to, and publishing, which describe the entity relationships in aviation equipment environmental testing.

[0050] Step S221: Construct a prompt word template for extracting entity relations from the aviation equipment environmental testing standard. The prompt words for extracting entity relations from the aviation equipment environmental testing standard mainly consist of the following parts: large language model task, entity extraction results, related domain entity relations, input aviation equipment environmental testing standard text, extraction example, and output format. The task module specifies that the large language model needs to refer to the example. Based on the entity extraction results and the relation setting list, extract all entity relations from the input national standard text and return the results according to the output format. The entity extraction result is the aviation equipment environmental testing standard entity sequence E obtained in step S21, the related domain relation setting list is the aviation equipment environmental testing standard entity relation RD obtained in step S13, and the input text of the aviation equipment environmental testing standard... Determined through step S11.

[0051] Step S222: Use a large language model to analyze the environmental testing standard text for aviation equipment. Entity relation extraction is performed; the objective of the relation extraction task is to extract the input text from the aviation equipment environmental testing standards. In the text section, input text segments are provided for each environmental testing standard for aviation equipment. Combined with the physical entities of the aviation equipment environmental testing standards included therein Determine the type of relationship between them. , here This indicates no relationship, meaning there is no relationship between the two aviation equipment environmental testing standard entities. Any relation within; relation extraction is a function mapping problem, specifically: ; in, For the entity relationship of environmental testing standards for aviation equipment; This indicates that there is no relation.

[0052] For a given input text segment of the aviation equipment environmental testing standard Physical counterpart to environmental testing standards for aviation equipment Extracting functions through entity relations Output of physical pairs of environmental testing standards for aviation equipment The types of relationships present in the text or output This indicates that there is no relation.

[0053] In implementation, the entity relationship of the aviation equipment environmental testing standard obtained in step S14 is used. The process of combining with templates to form prompt words is set as entity relation prompt words. The specific implementation is as follows: The "task" extracts the relationships between entities based on the provided "national standard text" and "annotated entities" and outputs the results according to the output format.

[0054] "Relationship Extraction Requirements" Core Relationships: Must include at least, but not limited to, the following relationship types: <Relationship Setting List> >

[0055] The “extraction principle” requires: accuracy: the relationship must be based on explicit textual description or strong implication and must not be fabricated; full coverage: all entity pairs that conform to the above relationship should be identified as much as possible; chained extraction: if A contains B and B contains C, then (A, contains, B) and (B, contains, C) need to be extracted simultaneously.

[0056] The output format is as follows: Each relation object must be output strictly in the format "AAA||XXX||BBB", where AAA represents the relation subject, XXX represents the relation type, and BBB represents the relation object.

[0057] In the extraction of entity relationships in the environmental testing standards for aviation equipment, entity relationship prompts will be used. Input text for environmental testing standards for aviation equipment List of entities for environmental testing standards for aviation equipment obtained in S21 The process of combining the input with a large language model is denoted as follows: And record the output result as Specifically: ; in, The results of entity relationship extraction for environmental testing standards of aviation equipment; This is a prompt for entity relationships.

[0058] When extracting entity relations, large language models use the following formula: ; in, List of entity relationships for environmental testing standards for aviation equipment.

[0059] For a given input text of the environmental testing standard for aviation equipment List of entities for environmental testing standards for aviation equipment and the predefined list of entity relationships This will yield a list of entity relationships with the highest probability. .in, This indicates how to generate a list of entity-relationships given the input text, entities, and constraints. The likelihood of the target. In implementation, the large language model will perform a simple comparison of each aviation equipment environmental testing standard entity pair. Calculate which relation type it belongs to. Based on the probability, select the relation type with the highest probability.

[0060] The example contains a standard input text for environmental testing of aviation equipment. Aircraft Equipment Environmental Testing Standard Input Text Segment The following is an example: 11. Information that should be provided in relevant specifications. When the relevant specifications include this test, the following information should be provided as much as possible; the severity level is temperature and the number of cycles. In this example, there are two entities related to the environmental testing standards for aviation equipment: the chapter entity "11. Information that should be provided in relevant specifications" and the item entity "a. Severity level: temperature and number of cycles," denoted as follows: and The information to be provided in the relevant specifications (section 11) includes the relationship between "a. Severity: Temperature and Cycle Count", denoted as... This is the input text segment of the aviation equipment environmental testing standard. and entity pair Specifically, "11. Information that should be given in the relevant specifications" and "a. Severity: temperature and number of cycles", extracted through a relational extraction function. , and thus "Contains". For other unrelated entity pairs, such as if the text also contains the entity "b, initial detection" (an aviation equipment environmental testing standard), let's set it as... ,So It may output This is because the relationship between "11. Information that should be given in the relevant specifications" and "b. Initial detection" is not explicitly mentioned in this text fragment.

[0061] Step S223: Parse the output of the large language model and perform the following transformations: ; This is the parsing function for relation recognition, used to ensure the consistency of the output results. In the parsing function for entity relation recognition... In this context, regular expressions are used for matching; the matching result is a triple of the form "AAA||XXX||BBB", which is the extracted result. Among them, the entity relationships of each aviation equipment environmental testing standard. For a triple: ; in, For the entity relationship triplet of the environmental testing standard for aviation equipment; It serves as the main entity for environmental testing standards for aviation equipment; For the entity relationship type of aviation equipment environmental testing standards; It refers to the physical entity that serves as the standard for environmental testing of aviation equipment.

[0062] Step S23: Construct a method for extracting entity attributes of aviation equipment environmental testing standards based on a large language model, and extract the entity attributes of aviation equipment environmental testing standards. The large language model is used to extract the entity attributes of aviation equipment environmental testing standards; in this step, the focus is on extracting physical quantity attributes related to the quantification of test conditions and results, including but not limited to the values ​​and units of dimensions, temperature, humidity, pressure, and duration.

[0063] Step S231: Construct an entity attribute extraction prompt template; the prompts for extracting entity attributes from the aviation equipment environmental testing standard mainly consist of the following parts: large language model task, entity extraction results, related domain entity attributes, input aviation equipment environmental testing standard text, extraction example, and output format. The task module specifies the large language model reference example. Based on the entity extraction results and attribute setting list, extract all aviation equipment environmental testing standard entity attributes from the input national standard text and return the results according to the output format. The entity extraction result is the aviation equipment environmental testing standard entity relation triplet obtained in step S22. The relevant domain attribute setting list is the entity attribute of the aviation equipment environmental testing standard obtained in step S14. Enter the text of the environmental testing standard for aviation equipment. This is obtained in step S11.

[0064] Step S232: Use a large language model to analyze the environmental testing standard text for aviation equipment. Perform entity attribute extraction; in attribute extraction, it is necessary to extract the entity attributes of the aviation equipment environmental testing standards obtained in step S14. The process of combining with templates to create entity attribute hints is set as the entity attribute hint function. The specific implementation is as follows: The "task" is to extract the attribute values ​​of entities based on the provided "national standard text" and "annotated entities" and output the results according to the output format.

[0065] In the "Extraction Scope and Entity Settings" section, please pay special attention to and extract the following entity types' attributes: <Attribute Settings List AD>.

[0066] Each attribute is output on a separate line in the format "Entity Name||Attribute Name||Attribute Value".

[0067] Entity attribute prompts Input text for environmental testing standards for aviation equipment The list of entities for environmental testing standards for aviation equipment obtained in step S21 The process of combining the input with a large language model is denoted as follows: And record the output result as Specifically: ; in, For physical attributes of environmental testing standards for aviation equipment.

[0068] Step S233: Parse the output of the large language model and perform the following transformations: ; in, List of entity attributes for environmental testing standards for aviation equipment; This is the parsing function for attribute recognition, used to ensure the consistency of the output results. (The rest of the text appears to be a fragment and requires further context for accurate translation.) In this context, regular expressions are used for matching; the matching result is a triple of the form "AAA||XXX||YYY", which is the extracted result. Among them, the physical attributes of each aviation equipment environmental test standard and It is a triple, specifically: ; in, For attribute entries; The main entity for which entity attributes are extracted; For entity attribute types; This corresponds to the value of the entity attribute.

[0069] Step S24: Based on the preliminary list of aviation equipment environmental testing standard entity types extracted in Step S21... Based on the entity relationship and attribute extraction results of the aviation equipment environmental testing standard, supplement the entity extraction results of the aviation equipment environmental testing standard.

[0070] Step S241: Traverse the entity relation extraction results of the aviation equipment environmental testing standard in step S22, count all subject entities and object entities, and check whether they appear in the entity extraction results. If they do not appear, record and save them; from the aviation equipment environmental testing standard entity relation set... Extract all main entities and object entity This forms a set of entities representing aviation equipment environmental testing standards in the relationship extraction results. for: ; in, This refers to the set of entities representing aviation equipment environmental testing standards in the relation extraction results. It serves as the main entity for environmental testing standards for aviation equipment; It refers to the physical entity that serves as the standard for environmental testing of aviation equipment.

[0071] For the list of entities that meet the environmental testing standards for aviation equipment The set of entities for environmental testing standards for aviation equipment in the relationship extraction results were examined. Does each aviation equipment environmental testing standard entity appear in the list? In the process, the first set of standard entities for environmental testing of aviation equipment was obtained. for: ; in, This is the first set of physical entities for environmental testing standards for aviation equipment. To include all in In but not in List of entity types for environmental testing standards for aviation equipment; This is a set difference operation.

[0072] Step S242: Traverse the entity attribute extraction results of aviation equipment environmental testing standards in step S23, count all main entities of aviation equipment environmental testing standards, and check whether they appear in the entity extraction results of aviation equipment environmental testing standards. If they do not appear, record and save them; from the aviation equipment environmental testing standard entity attribute set Extract the main entity of all aviation equipment environmental testing standards This forms a set of aviation equipment environmental testing standard entities in the attribute extraction results. : ; in, This refers to the set of entities representing aviation equipment environmental testing standards in the attribute extraction results.

[0073] For the list of entities that meet the environmental testing standards for aviation equipment Examine the set of aviation equipment environmental testing standard entities in the attribute extraction results. Does each aviation equipment environmental testing standard entity appear in the list? China; Second set of environmental testing standards for aviation equipment for: ; in, The second set of environmental testing standards for aviation equipment includes all those in... In but not in The physical entity of the environmental testing standards for aviation equipment.

[0074] Step S243: Combine the first set of aviation equipment environmental test standard entities obtained in steps S241 and S242. With the second set of environmental testing standards for aviation equipment The corresponding national standard text and the entity recognition results of the aviation equipment environmental testing standard are input into the large language model, combined with preset prompt words; the parts of the results saved in steps S241 and S242 that do not exist in the entity recognition results are filtered out, and the aviation equipment environmental testing standard entity list is updated for these parts of the results. .

[0075] Step S3: Detect aviation equipment environmental testing standard entity pairs with different expressions but the same meaning based on semantic vector similarity; merge those aviation equipment environmental testing standard entities with different textual expressions but the same semantics to ensure the uniqueness and consistency of aviation equipment environmental testing standard entities in the knowledge graph; specifically including the following sub-steps: Step S31: Standardize the list of standard entity types for environmental testing of aviation equipment; for the list of standard entities for environmental testing of aviation equipment obtained in step S21 The first set of standard entities for environmental testing of aviation equipment obtained in step S241 The second set of standard entities for environmental testing of aviation equipment obtained in step S242 Merge and update them to obtain the set of entities for environmental testing standards for aviation equipment. for: ; in, A collection of standard entities for environmental testing of aviation equipment. It includes all entities extracted from environmental testing standard documents; For the first A standard entity for environmental testing of aviation equipment, such as "GB / T2423.22-2012 / IEC60068-2-14:2009", where n is the number of entities, typically... .

[0076] This stage requires environmental testing standards for each piece of aviation equipment. Perform standardized cleaning, as follows: ; in, This refers to the pre-processed entity text of the environmental testing standard for aviation equipment, such as "gbt2423.22-2012 / iec60068-2-14:2009". In the implementation of the cleaning mode, The punctuation marks are set to "()" and "《》\s+" to remove unnecessary punctuation in the environmental testing standards for aviation equipment. To make the string Matching mode Partial replacement with ; Indicates the removal of the string. Leading and trailing whitespace characters, Indicates the string Converting to lowercase and then performing standardized cleaning yields a standardized set of entities. .

[0077] Step S32: Perform semantic vectorization of standard entities for aviation equipment environmental testing, and calculate the semantic similarity of standard entities for aviation equipment environmental testing; for the cleaned standardized entity set, use a transformer encoder to map the text to the semantic space as follows: ; in, For the first The original semantic vector of each aviation equipment environmental testing standard entity ; This is the encoder function for the Sentence-BERT model. This represents a 384-dimensional real vector space.

[0078] The semantic vectors of entities in the environmental testing standards for aviation equipment are L2 normalized, specifically as follows: ; in, The normalized semantic vectors of the entities in the environmental testing standards for aviation equipment form a set of normalized semantic vectors. ; For vectors The L2 norm modulus length; For vectors The The components of each dimension.

[0079] For the normalized aviation equipment environmental testing standard entity semantic vector set Calculate the semantic similarity matrix of entities in the environmental testing standards for aviation equipment. The calculation method is as follows: ; in, for The entity semantic similarity matrix; Entity semantic similarity matrix of OK The elements in the column position represent entities of environmental testing standards for aviation equipment. and Cosine similarity; semantic vector and The angle between them; The cosine of the included angle is equivalent to the similarity. .

[0080] Step S33: Perform pre-classification and screening of entities for aviation equipment environmental testing standards; obtain the entity semantic similarity matrix S and the set of entities for aviation equipment environmental testing standards. Using similarity thresholds For binary classification, the classification decision function is: ; ; in, This is a set of entity indexes for environmental testing standards of aviation equipment that need to participate in clustering. A set of indexes for independent and unique environmental testing standards for aviation equipment; The clustering threshold is set to 0.85 in the environmental testing standards for aviation equipment.

[0081] The clustered candidate aviation equipment environmental testing standard entity set is as follows The set of independent aviation equipment environmental testing standard entities is as follows: .

[0082] Perform entity semantic conflict detection for aviation equipment environmental testing standards, calculate entity semantic confidence of aviation equipment environmental testing standards, and obtain entity semantic similarity matrix S and aviation equipment environmental testing standard entity set. Detect conflict pairs that meet the following conditions: ; in, This is a set of semantic conflicts for standard entities in potential aviation equipment environmental testing. The similarity threshold is set at 0.85 in the environmental testing standard.

[0083] The confidence level of each entity semantic conflict pair in the environmental testing standards for aviation equipment is calculated as follows: ; in, For the semantic conflict of entities in the environmental testing standards for aviation equipment The detection confidence level.

[0084] The entity semantic conflict detection results for the aviation equipment environmental testing standard are as follows: ; in, The total number of semantic conflict pairs for entities in the environmental testing standards for aviation equipment.

[0085] Step S34: Semantic clustering analysis of entities for environmental testing of aviation equipment; This involves clustering the candidate sets of entities for environmental testing of aviation equipment obtained in step S33. The DBSCAN clustering algorithm is applied, specifically as follows: ; in, For the first The clustering label of each aviation equipment environmental test standard entity, when the value is -1, indicates that it is a noise point, that is, an independent entity that does not need to be clustered. DBSCAN is a density-based clustering algorithm function. The neighborhood radius parameter in the DBSCAN algorithm is used to determine whether two aviation equipment environmental test standard entities are "sufficiently similar". In the aviation equipment environmental test standard, it is set to 0.3. The minimum sample size parameter is set to 1.

[0086] The semantic clustering results of the entities in the environmental testing standards for aviation equipment were obtained as follows: for: ; in, The results of entity semantic clustering for environmental testing standards of aviation equipment; This refers to the semantic clustering category of entities in the environmental testing standards for aviation equipment.

[0087] Results integration and output; based on the results obtained in the above steps, statistical indicators are calculated; the output of this step mainly includes: entity semantic conflict pairs in aviation equipment environmental testing standards. ; Entity semantic clustering results of aviation equipment environmental testing standards Clustering candidate aviation equipment environmental testing standard entity set Independent aviation equipment environmental testing standard entity set The total number of entity semantic conflicts in the aviation equipment environmental testing standards is [number missing]. Average semantic similarity of entities in aviation equipment environmental testing standards Entity semantic clustering quality parameters for environmental testing standards of aviation equipment ;in, Semantic clustering categories for entities in aviation equipment environmental testing standards The internal average similarity.

[0088] Step S4: Consistency Detection of Aviation Equipment Environmental Testing Standard Entities Based on Graph Relationships and Attribute Similarity; This step aims to utilize the relational topology and attribute features of aviation equipment environmental testing standard entities in the knowledge graph to detect and merge those aviation equipment environmental testing standard entities with high similarity relationships in the graph, thus ensuring a consistent knowledge graph structure. The relational topology particularly focuses on the associations between the test equipment used in the test project, the characteristic test parameters of the test project, and the physical quantity attributes included in the test parameters; specifically, it includes the following sub-steps: Step S41: Extract the entity relationships and attribute sets of the aviation equipment environmental testing standards; based on the entity relationship extraction results of the aviation equipment environmental testing standards obtained in step S22. Its form is "Subject Entity || Relationship Type || Object Entity", and the entity attribute extraction results of the aviation equipment environmental test standard obtained in step S23 are as follows. Its form is "Main Entity || Attribute Type || Attribute Value". This applies to any two aviation equipment environmental testing standard entities to be compared. and Collect the entity relationship set of aviation equipment environmental testing standards respectively. , Set of physical attributes of aviation equipment environmental testing standards , ;in, Indicates the entity based on the environmental testing standards for aviation equipment. All relational triples that serve as either the subject or object entity. The entity relation for each aviation equipment environmental testing standard is represented as: Relationship Type, Associated Entity. This indicates that the entity includes environmental testing standards for aviation equipment. All attribute triples; each attribute is represented as: attribute type, attribute value; , Similarly.

[0089] Step S42: Calculate the entity relationship similarity of the aviation equipment environmental testing standards; use Jaccard similarity to calculate the entity relationship similarity of the aviation equipment environmental testing standards: ; in, Similarity of entity relationships in environmental testing standards for aviation equipment; This is the intersection of sets, i.e., common relation pairs; The union of sets is defined as all unique relation pairs. Two relation pairs are considered the same relation if they have the same relation type and associated entities.

[0090] Similarly, using Jaccard similarity to calculate the entity attribute similarity of the aviation equipment environmental testing standard, the following is also true: ; in, Similarity of physical attributes for environmental testing standards of aviation equipment; This is the intersection operator for sets; For the union of sets, two attribute pairs are considered the same if they have the same attribute type and attribute value.

[0091] A weighted average is calculated by combining the similarity of entity relationships and attributes in the environmental testing standards for aviation equipment, specifically as follows: ; in, For overall similarity, overall similarity The range is [0,1]. The closer the value is to 1, the more likely it is to be the same entity. α is the first weighting coefficient. β is the second weighting coefficient. The condition α+β=1 is met. In the environmental testing standard for aviation equipment, α=0.7 and β=0.3 are set.

[0092] Constructing a similarity matrix of entity graph relationships for environmental testing standards of aviation equipment for: ; in, for Similarity matrix of entity diagrams for environmental testing standards of aviation equipment; Graph Relationship Similarity Matrix of OK The elements in the column position represent entities of environmental testing standards for aviation equipment. and Graph similarity.

[0093] Step S43: Perform entity graph relationship conflict detection for aviation equipment environmental testing standards; based on the graph relationship similarity matrix, detect potentially conflicting entity pairs for aviation equipment environmental testing standards, specifically: ; in, For conflict aviation equipment environmental testing standard physical pair; The graph relationship similarity threshold is set to 0.85 in the environmental testing standard.

[0094] The confidence level for the conflict of entity diagram relationships in the environmental testing standards for aviation equipment is calculated as follows: ; in, Conflict confidence level of entity diagram relationships in environmental testing standards for aviation equipment; This is a function that takes the minimum value.

[0095] The results of the conflict detection of entity diagram relationships in the environmental testing standard for aviation equipment are as follows: ; in, The results of the entity diagram relationship conflict detection for environmental testing standards of aviation equipment.

[0096] Step S44: Perform entity clustering for aviation equipment environmental testing standards based on similarity thresholds; perform entity clustering directly using similarity thresholds based on the entity graph relationship similarity matrix of aviation equipment environmental testing standards. Initialize the entity clustering set for aviation equipment environmental testing standards: Clusters = ∅, and the unassigned entity set: U = { , ,..., }; For each unassigned aviation equipment environmental testing standard entity ∈U, create iterative clustering of standard entities for environmental testing of aviation equipment. ={ For each other unassigned aviation equipment environmental testing standard entity ∈U and ≠ If the similarity matrix of entity diagrams in the environmental testing standards for aviation equipment > Then Add iterative clustering of entities in aviation equipment environmental testing standards After completing one round of testing, remove the aviation equipment environmental test standard entities from the unassigned aviation equipment environmental test standard entity set U and perform iterative clustering. All entities in the aviation equipment environmental testing standard entity iterative clustering Join the Aviation Equipment Environmental Testing Standards Cluster; among them The threshold for graph relation clustering was set to 0.75; the resulting graph relation clustering results for unassigned aviation equipment environmental testing standards entities are as follows: ; in, Clustering results of entity graph relationships for unassigned aviation equipment environmental testing standards; This refers to the clustering categories of entity graph relationships for environmental testing standards of aviation equipment.

[0097] Based on the clustering results, the results are divided into: independent unassigned aviation equipment environmental testing standard entities and clustered unassigned aviation equipment environmental testing standard entities, specifically: ; ; in, List of independent, unassigned aviation equipment environmental testing standard entity types; This is a cluster of unassigned aviation equipment environmental testing standard entities.

[0098] Step S5: Merge the semantic vector similarity results of the aviation equipment environmental testing standard entities obtained in Step S3 with the relational similarity results of the aviation equipment environmental testing standard entity graphs obtained in Step S4: For each aviation equipment environmental testing standard entity semantic clustering ∈ Clustering of Entity Graph Relationships with Environmental Testing Standards for Aviation Equipment ∈ The similarity of entity clustering in the environmental testing standards for aviation equipment is calculated as follows: ; in, For the clustering similarity of entities in the environmental testing standards for aviation equipment; This is the result of semantic similarity clustering; The results of graph relationship similarity clustering.

[0099] For the clustering similarity of standard entities in environmental testing of aviation equipment Clustering, merging two clusters; for The clustering is retained, and the result of this process is recorded as the Entity Fusion Cluster Set of the Aviation Equipment Environmental Testing Standard. ; Determine the final list of entities for environmental testing standards for aviation equipment for: ; in, This is the final list of entities required for environmental testing standards for aviation equipment. A cluster set of entities for environmental testing standards for aviation equipment; It is a set of independent aviation equipment environmental testing standard entities obtained based on semantic similarity.

[0100] Final list of entities for environmental testing standards for aviation equipment Entity Relationship Extraction Results of Aviation Equipment Environmental Testing Standards The results of entity attribute extraction for aviation equipment environmental testing standards obtained in step S23 The entities, relationships, and attributes extracted by this invention together constitute the core framework of the knowledge graph of environmental testing standards for aviation equipment. Specifically, entities serve as graph nodes, relationships as edges, and attributes as node features, forming a complete graph structure data, namely the knowledge graph of environmental testing standards for aviation equipment.

[0101] The semantic consistency-based knowledge graph construction method for aviation equipment environmental testing proposed in this invention has broad and profound practical application value in the engineering field. Its core value lies in transforming massive, unstructured aviation equipment environmental testing standard texts into a structured, computable, and reasonable knowledge network, thereby directly empowering core industrial processes such as R&D design, manufacturing, quality inspection, and equipment operation and maintenance.

[0102] In the implementation examples, in the research and development and manufacturing of high-end complex equipment, such as aerospace, rail transportation, and smart grid equipment, this method can construct a professional knowledge graph covering a series of standards such as GB / T and GJB. Specifically, in the research and development of aero-engines, a knowledge graph of environmental testing standards for aerospace equipment is constructed, integrating numerous standards such as high-temperature testing, vibration testing, and fatigue testing. This allows for intelligent retrieval of all test items, qualification criteria, and equipment requirements related to current design parameters, greatly improving research and development efficiency and standard compliance. Furthermore, in the upgrading of intelligent testing and inspection equipment, the knowledge graph constructed by this method is embedded into the equipment control system, giving it the ability to "understand" and "execute" standards. For example, based on the constructed GB / T2423 series standard graph, the environmental test chamber automatically matches and sets corresponding temperature, humidity, vibration, and other test conditions and severity levels according to the type of product under test, realizing the intelligentization and standardization of the testing process, thereby improving the objectivity of test results and analysis efficiency.

[0103] To verify the effectiveness and reliability of the semantic consistency-based knowledge graph construction method for aviation equipment environmental testing proposed in this application, an evaluation experiment was designed. The following sections will quantitatively analyze the results of entity extraction, relation extraction, and attribute extraction using metrics such as precision, recall, and F1-score to objectively evaluate the method's performance in practical applications.

[0104] To ensure the reliability and validity of the evaluation results, this invention first manually annotates the entities and relationships of the aviation equipment environmental testing standards in multiple national standard documents according to established entity recognition and relationship extraction rules, constructing a high-quality annotated dataset, and using this annotated dataset as the benchmark for evaluation.

[0105] This method and the comparison method statistically analyzed the number of TruePositive (TP), FalseNegative (FN), and FalsePositive (FP) outputs. TP refers to the number of triples that exist in the manually labeled file and are correctly labeled in the output file, reflecting the valid entity relations accurately identified by the model; FN represents the number of triples that exist in the manually labeled file but not in the output file, reflecting missed detections of actual relations by the model; FP represents the number of triples that exist in the output file but not in the manually labeled file, i.e., entity relations incorrectly identified by the model. For example: The following entity exists in the manually annotated file: "GB / T8170||Standard; Preface||Chapter"; the output result is: "GB / T8170||Standard 3 Terms and Definitions||Chapter". It can be seen that the entity "GB / T8170||Standard" appears in both the manually annotated file and the output file, so it is considered a TP example. The entity "Preface||Chapter" exists in the manually annotated file but does not exist in the output file, so it is considered an FN example. The entity "3 Terms and Definitions||Chapter" exists in the output file but does not exist in the manually annotated file, so it is considered an FP example.

[0106] The following is a detailed introduction to each evaluation indicator: Precision measures the accuracy of a model in named entity recognition and relation extraction; it represents the proportion of correctly predicted values ​​out of all output text. Mathematically, precision is calculated as the ratio of the number of true positives (TP) to the sum of the number of true positives (TP) and false positives (FP). The formula is: ; in, Precision is the ratio of the number of true positives (TP) to the sum of the number of true positives (TP) and the number of false positives (FP). The higher the precision, the more accurate and reliable the model's identification results are when recognizing specific relationships. This represents the number of triples that exist in the manually annotated file and are also correctly annotated in the output file. This represents the number of triples that exist in the output file but not in the manually annotated file.

[0107] Recall focuses on evaluating the comprehensiveness of named entity recognition and relation extraction, that is, the proportion of correct parts in the output content to all correct content in the standard file; it is calculated as the ratio of the number of true positives (TP) to the sum of the number of true positives (TP) and the number of false negatives (FN), as shown in the following formula: ; in, Recall is the ratio of the number of true positives (TP) to the sum of the number of true positives (TP) and the number of false negatives (FN). A higher recall means that the model can identify more real relationships, has a higher degree of coverage of real relationships, and has fewer missed detections. This represents the number of triples that exist in the manually annotated file but not in the output file.

[0108] The F1 score, a harmonic mean of precision and recall, is a comprehensive metric that integrates these two metrics. It balances the relationship between precision and recall, avoiding the limitations of evaluating a single indicator. The F1 score is calculated as the ratio of the product of precision and recall to twice the sum of precision and recall. The formula is as follows: ; in, The F1 score is the ratio of the product of precision and recall to twice the sum of precision and recall. A higher F1 score indicates that the model performs well in both precision and recall, and has better overall performance. Introducing the F1 score allows for a more comprehensive and accurate evaluation of the overall performance of different methods in entity recognition and relation extraction tasks.

[0109] In the evaluation experiments, this study selected GraphRAG and LightRAG, both knowledge graph construction methods, as comparison objects. GraphRAG is an architecture that utilizes knowledge graph structures to enhance the reasoning capabilities of large language models. It constructs a graph structure from text and uses it for global information retrieval, but typically relies on predefined or general-domain entity and relation schemas. LightRAG, on the other hand, is a lightweight retrieval enhancement generation method that focuses on obtaining relevant context through efficient methods such as vector retrieval, but it is weaker in the deep parsing and structuring of complex semantic relationships. Figure 6 As shown in the specific evaluation stage of the relation extraction task, since the storage method of GraphRAG is relatively complex and it is difficult to directly represent the relation extraction results using triples, this method is only compared with LightRAG in terms of relation extraction.

[0110] Table 3 Performance evaluation of the three named entity recognition methods This method 92.35% 83.2% 87.54% GraphRAG 20.45% 19.07% 19.74% LightRAG 30.49% 28.81% 29.63% As shown in Table 3, in the named entity recognition task, our method outperforms GraphRAG and LightRAG in terms of precision, recall, and F1 score. This result fully demonstrates that our method can accurately identify entities in text in the named entity recognition task, while having fewer false negatives and missed detections, and its overall performance is superior to GraphRAG and LightRAG.

[0111] Table 4. Relationship between this method and LightRAG: Extraction performance evaluation This method 90.51% 82.48% 86.31% LightRAG 25.73% 32.23% 28.61% As shown in Table 4, this method also demonstrates excellent performance in the relation extraction task. It outperforms LightRAG in precision, recall, and F1 score. This result indicates that in the relation extraction task, this method can more accurately identify the relationships between standard entities in aviation equipment environmental testing, effectively reducing false positives and false negatives, and its performance is superior to LightRAG.

[0112] The second aspect of this invention proposes a knowledge graph construction system based on a semantic consistency-based method for constructing a knowledge graph for environmental testing of aviation equipment. The system includes an entity recognition module, a relationship recognition module, an attribute recognition module, and a data processing module.

[0113] The entity recognition module is responsible for automatically identifying and extracting a list of entity types from the preprocessed aviation equipment environmental test standard text. Specifically, this includes: constructing a prompt template for entity extraction, which contains a task description for the large language model, a list of aviation equipment environmental test standard entity types, input text, extraction examples, and output format; using the large language model to extract aviation equipment environmental test standard entities from the input text, extracting all eligible entities based on the given list of entity types, and outputting them in the format "entity name || entity type"; parsing the model output, ensuring consistent output format through regular expression matching, removing irrelevant text, and generating a list of aviation equipment environmental test standard entities, where each entity includes its name and type; and achieving automated identification of key entities in the aviation equipment environmental test standard text, providing foundational data for subsequent relationship and attribute identification.

[0114] The relation recognition module is responsible for extracting semantic relationships between aviation equipment environmental testing standard entities from the aviation equipment environmental testing standard text and the identified aviation equipment environmental testing standard entities. Specific functions include: constructing relation extraction prompt templates, which contain a task description from a large language model, the aviation equipment environmental testing standard entity extraction results, a relation setting list from relevant domains, input text, extraction examples, and output format; using a large language model to extract relations from the input text and entity list, the model determines whether a pre-defined relationship exists between pairs of aviation equipment environmental testing standard entity pairs based on the relation setting list, and outputs the result in the format "subject entity || relation type || object entity"; parsing the model output to ensure format consistency, and generating a list of aviation equipment environmental testing standard entity relations, where each aviation equipment environmental testing standard entity relation includes a subject entity, a relation type, and an object entity; this module establishes associations between aviation equipment environmental testing standard entities, forming relation edges in a knowledge graph, enhancing the structured representation of knowledge.

[0115] The attribute recognition module is responsible for extracting attribute information of aviation equipment environmental testing standard entities from the text of aviation equipment environmental testing standards and the identified aviation equipment environmental testing standard entities. Specific functions include: constructing attribute extraction prompt templates, which contain a task description of the large language model, the extraction results of aviation equipment environmental testing standard entities, a list of attribute settings in relevant fields, input text, extraction examples, and output format; using the large language model to extract attributes from the input text and the list of aviation equipment environmental testing standard entities; extracting corresponding attribute values ​​for each aviation equipment environmental testing standard entity based on the attribute list, and outputting the results in the format "entity||attribute type||attribute value"; parsing the model output to ensure format consistency and generating an attribute list, where each attribute includes the aviation equipment environmental testing standard entity, attribute type, and attribute value; this module enriches the detailed information of aviation equipment environmental testing standard entities and improves the feature descriptions of aviation equipment environmental testing standard entities in the knowledge graph, thereby supporting more granular knowledge query and analysis.

[0116] The data processing module is responsible for integrating, cleaning, and semantically consistent optimizing the entity, relation, and attribute data of aviation equipment environmental testing standards output by the aforementioned modules. Specific functions include: traversing the relation extraction and attribute extraction results, selecting aviation equipment environmental testing standard entities not present in the entity extraction results, and adding them as new aviation equipment environmental testing standard entities. Based on semantic vector similarity and graph relation and attribute feature similarity, it detects aviation equipment environmental testing standard entity pairs that have "different expressions but the same meaning" and performs semantic clustering on them. This module performs post-processing on the preliminary extraction results, ensuring high accuracy and consistency in the knowledge graph of environmental testing aviation equipment standards.

[0117] The beneficial effects of this invention are as follows: This invention constructs a knowledge graph of aviation equipment environmental testing standards using a large language model. Addressing the lack of dynamic adaptation in the knowledge system of aviation equipment environmental testing standards, it dynamically constructs a knowledge system library of entity types, relationships, and attribute settings for aviation equipment environmental testing projects, equipment, and parameters using a large language model. This enables targeted guidance for aviation equipment testing, improves the accuracy and generalization ability of knowledge extraction, and reduces the cost of manual annotation and reliance on feature engineering. Collaborative extraction is performed through a unified large language model framework, and customized prompt word templates are used to achieve high-precision information extraction, ensuring semantic consistency and integrity among entities, relationships, and attributes of aviation equipment environmental testing projects, equipment, and parameters, reducing data redundancy and omissions. Regarding redundancy and alignment issues caused by synonymous entities, traversal matching and semantic disambiguation using a large language model are used to verify and merge entities of unmatched aviation equipment environmental testing projects, equipment, and parameters, effectively improving the alignment quality and overall consistency of the knowledge graph and supporting more efficient knowledge reasoning and application.

[0118] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for constructing a knowledge graph for environmental testing of aviation equipment based on semantic consistency, characterized in that: It includes: S1: Optical character recognition of aviation equipment environmental testing standard documents yields the input text of aviation equipment environmental testing standards; Pre-defined entity types, entity relationships, and entity attributes for aviation equipment environmental testing standards; aviation equipment environmental testing standard entities include: test items, test equipment, and test parameters; constructing a knowledge system base for aviation equipment environmental testing standards; S2: Construct a standard entity extraction method for aviation equipment environmental testing using a large language model, extract the standard entities, entity relationships, and entity attributes of aviation equipment environmental testing; supplement the standard entity extraction results for aviation equipment environmental testing based on the entity relationship and entity attribute extraction results; S3: Perform semantic vectorization of standard entities for environmental testing of aviation equipment, calculate semantic similarity of standard entities for environmental testing of aviation equipment, and classify and filter entities; use semantic vector similarity to detect standard entity pairs for environmental testing of aviation equipment, and merge standard entities for environmental testing of aviation equipment with the same semantics. Step S3 is as follows: S31: Standardize the processing of environmental testing standard entities for aviation equipment to obtain a set of environmental testing standard entities for aviation equipment. , , For the first One entity for environmental testing standards for aviation equipment; S32: Perform semantic vectorization of entities in the aviation equipment environmental testing standards, and calculate the semantic similarity matrix of entities in the aviation equipment environmental testing standards. ; S33: Perform pre-classification and screening of standard entities for environmental testing of aviation equipment; obtain semantic conflict detection results for standard entities for environmental testing of aviation equipment. ; S34: Semantic clustering analysis of entities in aviation equipment environmental testing standards yielded the following results: Output semantic vector similarity detection results; S4: Based on graph relationships and attribute similarity, achieve consistency detection of aviation equipment environmental test standard entities; utilize the relational topology and attribute characteristics of aviation equipment environmental test standard entities in the knowledge graph to determine the association relationships of test equipment used in test projects, test parameters of test projects, and test parameters containing physical attribute characteristics; detect and merge aviation equipment environmental test standard entities with similar relationships to make the knowledge graph structure consistent. Step S4 is as follows: S41: Entity relationship extraction results based on aviation equipment environmental testing standards Extraction results of entity attributes of environmental testing standards for aviation equipment Extract the entity relationships and attribute sets of aviation equipment environmental testing standards; S42: Calculate the similarity of entity relationships in environmental testing standards for aviation equipment Similarity of attributes The overall similarity is obtained by weighted averaging. Construct a similarity matrix of entity graph relationships for environmental testing standards of aviation equipment. ; S43: Conduct conflict detection of relationship in the entity diagram of the environmental testing standard for aviation equipment; calculate the confidence level of conflict in the entity diagram of the environmental testing standard for aviation equipment. The results of the entity diagram relationship conflict detection for the environmental testing standard of aviation equipment were obtained. ; S44: Cluster the entities of the aviation equipment environmental testing standards based on the similarity threshold to obtain independent unassigned aviation equipment environmental testing standard entities. Clustering of unassigned aviation equipment environmental testing standard entities ; S5: Calculate the cluster similarity of entities in the aviation equipment environmental testing standard; fuse the semantic vector similarity results of the aviation equipment environmental testing standard entities in step S3 with the entity graph relationship similarity results of the aviation equipment environmental testing standard in step S4; determine the final list of aviation equipment environmental testing standard entities and obtain the aviation equipment environmental testing standard knowledge graph. Step S5 is as follows: The similarity of entity clustering in the environmental testing standards for aviation equipment is calculated as follows: ； in, For the clustering similarity of entities in the environmental testing standards for aviation equipment; This is the result of semantic similarity clustering; The results of graph relationship similarity clustering; For the semantic clustering categories of entities in the environmental testing standards for aviation equipment; Clustering categories for entity graph relationships in aviation equipment environmental testing standards; Based on the entity clustering similarity of aviation equipment environmental testing standards Cluster extraction was performed to obtain the entity fusion cluster set of aviation equipment environmental testing standards. ; Determine the final list of entities for environmental testing standards for aviation equipment .

2. The method for constructing a knowledge graph for aviation equipment environmental testing based on semantic consistency according to claim 1, characterized in that: Step S33 is as follows: Perform pre-classification and screening of entities for environmental testing standards of aviation equipment; obtain the entity semantic similarity matrix S and the set of entities for environmental testing standards of aviation equipment. Using similarity thresholds Binary classification was performed to obtain the clustered candidate aviation equipment environmental testing standard entity set as follows: and the set of independent aviation equipment environmental testing standards entities ; This is a set of entity indexes for environmental testing standards of aviation equipment that need to participate in clustering. This involves creating an index set of independent and unique standard entities for aviation equipment environmental testing; performing semantic conflict detection on these entities; calculating their semantic confidence; and obtaining the entity semantic similarity matrix S and the set of standard entities for aviation equipment environmental testing. The detection yielded a set of semantic conflicts for entities that meet the environmental testing standards for aviation equipment. ; Calculate the confidence level of each entity semantic conflict pair in the environmental testing standards for aviation equipment. ; The results of entity semantic conflict detection for aviation equipment environmental testing standards were obtained as follows: .

3. The method for constructing a knowledge graph for aviation equipment environmental testing based on semantic consistency according to claim 1, characterized in that: Step S34 outputs the semantic vector similarity detection results, specifically including: Semantic conflicts of entities in aviation equipment environmental testing standards Semantic clustering results of aviation equipment environmental testing standards Clustering candidate aviation equipment environmental testing standard entity set Independent aviation equipment environmental testing standard entity set The total number of entity semantic conflicts in the environmental testing standards for aviation equipment is [number missing]. Average semantic similarity of entities in environmental testing standards for aviation equipment Entity semantic clustering quality parameters for environmental testing standards of aviation equipment ,in, Semantic clustering categories for entities in aviation equipment environmental testing standards The internal average similarity.

4. The method for constructing a knowledge graph for aviation equipment environmental testing based on semantic consistency according to claim 1, characterized in that: Step S42 is as follows: Using Jaccard similarity to calculate entity relation similarity and attribute similarity in aviation equipment environmental testing standards: ； ； in, For the similarity of entity relationships in environmental testing standards for aviation equipment; This is the intersection operator for sets; This is the set union operator; Similarity of physical attributes for environmental testing standards of aviation equipment; A weighted average is calculated by combining the similarity of entity relationships and attributes in the environmental testing standards for aviation equipment, specifically as follows: ； in, α represents the overall similarity; α is the first weighting coefficient; β is the second weighting coefficient.

5. The method for constructing a knowledge graph for aviation equipment environmental testing based on semantic consistency according to claim 1, characterized in that: Step S2 is as follows: S21: Extract entities from the aviation equipment environmental testing standards using a large language model, and output a list of entity types in the aviation equipment environmental testing standards that includes entity names and entity types. This includes: test items, test equipment, and test parameters; S22: Use large language models to extract entity relationships in aviation equipment environmental testing standards, including the relationship types of test items, test equipment, and test parameters; S23: Use large language models to extract the entity attributes of aviation equipment environmental testing standards, and extract physical quantity attributes related to test conditions, including test equipment size, test temperature, and test humidity; S24: List of standard entity types for aviation equipment environmental testing initially extracted in step S21 Based on the entity relationship and attribute extraction results of the aviation equipment environmental testing standard, supplement the entity extraction results of the aviation equipment environmental testing standard.

6. The method for constructing an aviation equipment environmental testing knowledge graph based on semantic consistency according to claim 1, characterized in that: Step S1 is as follows: S11: Obtain the aviation equipment environmental testing standard document in portable PDF format, perform optical character recognition, and obtain the original recognized aviation equipment environmental testing standard text. ; S12: List of entity types for pre-defined environmental testing standards for aviation equipment; S13: Pre-defined entity relationships for environmental testing standards of aviation equipment; S14: Pre-defined entity attributes for environmental testing standards of aviation equipment; S15: List of Entity Types for Environmental Testing Standards for Aviation Equipment obtained from steps S12-S14 Entity Relationship and entity attributes To build a knowledge system database of environmental testing standards for aviation equipment.

7. A knowledge graph construction system for the knowledge graph construction method based on semantic consistency for aviation equipment environmental testing as described in any one of claims 1 to 6, characterized in that, It includes: Entity recognition module, relationship recognition module, attribute recognition module, and data processing module; The entity recognition module extracts the prompt word template of the aviation equipment environmental test standard entity, uses a large language model to extract the aviation equipment environmental test standard entity from the input text, extracts all aviation equipment environmental test standard entities that meet the conditions, generates a list of aviation equipment environmental test standard entities, and realizes the automatic recognition of key entities in the aviation equipment environmental test standard text. The relationship recognition module establishes the associations between entities in the aviation equipment environmental testing standard, forming relationship edges in the knowledge graph, enhancing the structured representation of knowledge, and extracting semantic relationships between entities in the aviation equipment environmental testing standard from the aviation equipment environmental testing standard text and the identified aviation equipment environmental testing standard entities. The attribute recognition module includes detailed information on the entities of the aviation equipment environmental testing standards, improves the feature descriptions of the entities of the aviation equipment environmental testing standards in the knowledge graph, thereby supporting more granular knowledge query and analysis, and is responsible for extracting the attribute information of the aviation equipment environmental testing standards entities from the aviation equipment environmental testing standard text and the identified aviation equipment environmental testing standard entities. The data processing module is responsible for integrating, cleaning, and semantically optimizing the entity, relation, and attribute data of the aviation equipment environmental testing standards output by the aforementioned modules.

Images