Construction method and system of intelligent mine key technology knowledge graph

Abstract

The invention provides an intelligent mine key common technology knowledge graph construction method and system. The method comprises the following steps: constructing an intelligent mine key technology related data set; data features are analyzed, and an intelligent mine construction and operation key generality technology ontology model with the key technology-application scene-dependent facility as the core is constructed; performing entity recognition and relation extraction on the unstructured data by adopting a deep learning model fusing a BERT pre-training language model and a Lattice-LSTM structure; and obtaining a visual entity relationship triple and storing the visual entity relationship triple in a graph database. According to the application, by promoting fusion of the knowledge graph and the intelligent mine field, the challenge of knowledge fragmentation can be solved, and the constructed knowledge graph can optimize the current intelligent mine key technology framework and analyze the development path of the intelligent mine key technology, so that theoretical and technical support is provided for intelligent transformation and upgrading of traditional mining operation.

Description

Technical Field

[0001] The present application relates to the technical field of cross-application of smart mines and knowledge graphs, and specifically to a method and system for constructing a knowledge graph of key technologies in smart mines. Background Art

[0002] In recent years, the mining industry has achieved remarkable success in construction and operations driven by informatization, digitalization, and intelligence. By introducing cutting-edge technologies such as the Internet of Things, big data, and artificial intelligence, the automation level and decision-making efficiency of mining production processes have been significantly improved. Breakthroughs have also been made in areas such as resource optimization, production safety management, and environmental protection. However, with the rapid development of technology, emerging scientific and technological advancements in the mining sector are constantly being explored and applied. Their inherent complexity, interdisciplinary nature, and iterative evolution have resulted in relevant knowledge being dispersed across vast amounts of academic literature, patents, and technical reports. This fragmentation not only increases the difficulty of knowledge acquisition and integration but also easily leads to the emergence of information silos, posing a significant challenge to the comprehensive and efficient analysis of technological development paths and innovation directions. Specifically, technical knowledge in the mining sector spans multiple disciplines, including geological exploration, mining engineering, equipment operation and maintenance, and environmental monitoring, with these different technologies intertwined and influencing each other. Rapid technological evolution also leads to frequent knowledge updates. While traditional databases can handle structured data, they lack the ability to handle unstructured data, semantic associations, and dynamic updates. Therefore, there is an urgent need for a new knowledge management method that can support multi-source heterogeneous data integration, semantic association analysis and dynamic knowledge updating to effectively store, integrate and utilize technical information.

[0003] Knowledge graphs (KGs), a knowledge representation method organized in the form of semantic networks, present information graphically, and intuitively display the relationships between research objects. They are widely used in fields such as search engines, recommender systems, question-answering systems, intelligent customer service, and natural language processing. As an emerging branch of artificial intelligence, knowledge graphs systematically process massive amounts of internet data and transform it into structured knowledge. The construction of a knowledge graph primarily involves three key processes: knowledge extraction, knowledge fusion, and knowledge graph construction. Knowledge extraction involves automatically identifying and extracting entities and their relationships from text corpora; knowledge fusion integrates the extracted knowledge, eliminates redundant information, and processes entity and relationship attributes to ensure consistency and accuracy; and knowledge graph construction involves storing the fused knowledge in a graph database, ultimately forming a visual knowledge graph. Existing knowledge graph construction methods are primarily divided into rule-based and machine learning-based approaches. Rule-based approaches require manual definition of entity and relationship rules, resulting in a complex construction process and limitations in scalability and adaptability. The machine learning-based methods mainly include the construction and training of models such as long short-term memory networks (LSTM), conditional random fields (CRF), and hidden Markov models (HMM). These models perform well in object recognition, but still require processing steps such as word segmentation and disambiguation to improve accuracy.

[0004] In the mining field, domestic and foreign scholars have begun to explore the core role of knowledge graphs in promoting the development of intelligent mines, and its application scenarios have covered key links such as safe production, equipment operation and maintenance, resource optimization, and ecological governance. For example, by integrating geological exploration and equipment operation data through knowledge graphs, mining paths can be optimized and resource utilization efficiency can be improved; building a mine safety risk assessment model based on knowledge graphs can achieve hidden danger prediction and accident warning; combining knowledge graphs with remote sensing technology can support the dynamic generation of ecological restoration plans. These practices show that knowledge graphs solve the pain points of information dispersion and decision lag in certain links of traditional mines through semantic association and fusion of multi-source heterogeneous data, and promote the digital transformation of mines. However, existing research still has the following limitations, (1) The key technology system for the entire life cycle of a mine still lacks systematic integration. Traditional technology hotspot analysis ignores the complex relationships and update laws between technologies, and the research results cannot support in-depth technology development analysis. (2) Existing knowledge graphs related to smart mines mostly focus on a single scenario. The semantic alignment efficiency and accuracy of multi-source data in the knowledge graph construction process are insufficient, and the knowledge associations between core technologies such as geological modeling, intelligent sensing, and digital twins have not been fully explored; (3) The entity extraction and relationship extraction models do not sufficiently mine the text features in the mining field, resulting in low recognition accuracy and poor recognition of entity overlap. Specifically, in terms of entity recognition, the text in the specific field of intelligent mining has a large number of proper nouns and syntactic features, and the field features need to be taken into account in the optimization of recognition effects. Summary of the Invention

[0005] This application provides a method and system for constructing a knowledge graph of key technologies in intelligent mines, which can solve the technical problems of knowledge fragmentation, lack of semantic association and insufficient cross-domain technology integration existing in the existing technology.

[0006] In the first aspect, a method for constructing a knowledge graph of key common technologies for intelligent mines includes the following steps: Step S1: Collect key common technical data on construction and operation from relevant databases in the field of smart mines to construct a data set; Step S2: Analyze and extract the characteristic information of the data set, and construct an ontology model of key common technologies for intelligent mine construction and operation containing a multi-level classification system; Step S3: Based on the constructed key common technology ontology model for intelligent mine construction and operation, a deep learning model integrating the BERT pre-trained language model and the Lattice-LSTM structure is used to perform entity recognition and relationship extraction on unstructured data to obtain entity relationship triples of entities, relationships, and attributes; Step S4: Visualize the entity relationship triple data and store it in a graph database to obtain a knowledge graph of key technologies for intelligent mines.

[0007] Furthermore, the step S1: collecting key common technical data on construction and operation from relevant databases in the field of intelligent mines to construct a data set specifically includes the following steps: Build a technical knowledge graph database in the field of intelligent mining; The constructed technical knowledge graph database in the field of intelligent mining is preprocessed to construct a data set.

[0008] Furthermore, the step S2: analyzing and extracting the characteristic information of the data set and constructing a key common technology ontology model for intelligent mine construction and operation including a multi-level classification system specifically includes the following steps: Extract data features from the dataset; Establish a three-level core concept classification system of "key technology - application scenario - dependent facilities" based on the extracted data features; Based on the established three-level core concept classification system, a relationship system between concepts is constructed, and the attribute system and corresponding rules of each concept are defined to construct a key common technology ontology model for intelligent mine construction and operation that includes a multi-level classification system.

[0009] Furthermore, the specific composition of the ontology model of key common technologies for intelligent mine construction and operation includes concept class, attribute class, relationship class and rule class; among them, the concept class includes three core categories: key technologies, application scenarios and dependent facilities, and each category has subdivided sub-concepts; the attribute class is used to describe the specific characteristic parameters of each concept, including technical methods, system optimization goals and facility deployment costs; the relationship class defines various association methods between concepts, including technical dependency relationships, equipment collaborative relationships, etc.; the rule class includes constraints on various entities and relationships between entities, as well as logical reasoning rules, which together constitute a complete ontology model system.

[0010] Furthermore, step S3: based on the constructed key common technology ontology model for intelligent mine construction and operation, a deep learning model integrating the BERT pre-trained language model and the Lattice-LSTM structure is used to extract knowledge from unstructured data to obtain entity relationship triple data of entities, relationships and attributes, specifically including the following steps: Based on the constructed ontology model of key common technologies for intelligent mine construction and operation, the BERT encoding layer and Lattice-LSTM structure network are used to perform entity recognition and relationship extraction on the preprocessed data to obtain entity relationship triple data of entities, relationships and attributes; Extract triplet data through attention mechanism and CFR layer decoding optimization, and calculate loss to train the ontology model until the model converges to obtain optimized triplet data; Entity disambiguation and entity fusion are performed on the optimized triple data to generate triple data after entity disambiguation and entity fusion.

[0011] Furthermore, the step S4 visualizes the entity relationship triple data and stores it in a graph database to obtain a knowledge graph of key technologies of intelligent mines, which specifically includes the following steps: The optimized triple data is converted into graph database nodes and edges according to the preset mapping rules. Key technology entities, application scenario entities, and dependent facility entities correspond to different types of nodes, and the relationships between these entities are converted into connecting edges. The visualization function of the graph database is used to achieve an intuitive display of the knowledge graph. At the same time, an efficient query channel is established through the database indexing mechanism to complete the structured storage and visual presentation of key technical knowledge of intelligent mines.

[0012] Furthermore, the visualization function of the graph database is used to realize the visualization of the knowledge graph, which specifically includes the following steps: The concepts in the mapping ontology model are nodes, the mapping relationships are edges, and additional attributes are added to the corresponding nodes. The D3.js visualization framework is used to visualize the data into a knowledge graph and store it in the Neo4j graph database.

[0013] Secondly, this application provides a system for constructing a knowledge graph of key common technologies for intelligent mines, including: The data acquisition and preprocessing module is used to collect key common technical data on construction and operation from relevant databases in the field of intelligent mining and construct a data set; An ontology model building module is used to analyze and extract characteristic information from data sets and construct an ontology model of key common technologies for intelligent mine construction and operation, which includes a multi-level classification system. The knowledge extraction module is used to perform entity recognition and relationship extraction on unstructured data based on the constructed key common technology ontology model for intelligent mine construction and operation, using a deep learning model that integrates the BERT pre-trained language model and the Lattice-LSTM structure to obtain entity-relationship triple data of entities, relationships, and attributes; The knowledge storage and visualization module is used to visualize the entity relationship triple data and store it in a graph database to obtain a knowledge graph of key technologies in intelligent mines.

[0014] Furthermore, the ontology model building module includes: A feature extraction unit, used to extract data features from a data set; A concept classification system establishment unit, in communication with the feature extraction unit, for establishing a three-level core concept classification system of "key technology - application scenario - dependent facilities" based on the extracted data features; The ontology model construction unit, based on the three-level core concept classification system, constructs an association system between concepts, defines the attribute system of each concept and the corresponding rules, and constructs an ontology model of key common technologies for intelligent mine construction and operation that includes a multi-level classification system.

[0015] In a third aspect, the present application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the method for constructing a knowledge graph of key common technologies for intelligent mines as described above.

[0016] The beneficial effects of the technical solutions provided in the embodiments of the present application include at least: This application innovatively applies knowledge graph technology to the research of key technologies for intelligent mine construction and operation. Based on unstructured corpus, it helps overcome the bottlenecks of intelligent mine technology innovation and system application by extracting, integrating, processing and utilizing the numerous entities and relationships embedded in key technologies. Proposing strategies and application methods for building key technology knowledge graphs suitable for the industry context can clarify the relevance, development paths, and innovation laws of technologies in the construction and operation of smart mines, while also providing guidance and assistance to scholars and companies in related fields in understanding technology needs and development directions. By constructing an ontology of key technologies in intelligent mining and using the BERT-Lattice-LSTM model to extract named entities and entity relationships, the gap in the knowledge graph of key technologies in intelligent mining is filled. By identifying named entities in the field of mining technology and semi-automatically extracting unstructured knowledge based on the domain ontology, the visualization of entity interaction relationships and knowledge reasoning are realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A flowchart of a method for constructing a knowledge graph of key technologies for intelligent mines provided in an embodiment of the present application; Figure 2 A flowchart for constructing a knowledge graph of key technologies for intelligent mines provided in the embodiments of this application; Figure 3 A partial structural diagram of the key common technology ontology model for intelligent mine construction and operation provided in the embodiments of this application; Figure 4 A schematic diagram of the structure of the BERT-Lattice-LSTM model provided in an embodiment of the present application; Figure 5 A knowledge graph constructed based on key technologies provided in the embodiments of this application. DETAILED DESCRIPTION

[0018] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0019] The terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices. The terms "first", "second" and "third" are used to distinguish different objects, etc., and do not represent a sequence, nor do they limit the "first", "second" and "third" to different types.

[0020] In the description of the embodiments of this application, the words "exemplary," "for example," or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary," "for example," or "for example" in the embodiments of this application should not be construed as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary," "for example," or "for example" is intended to present the relevant concepts in a concrete manner.

[0021] In the description of the embodiments of the present application, unless otherwise specified, “ / ” means or, for example, A / B can mean A or B; “and / or” in the text is merely a description of the association relationship of associated objects, indicating that three relationships may exist, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, “multiple” refers to two or more than two.

[0022] In some processes described in the embodiments of the present application, multiple operations or steps are included that appear in a specific order. However, it should be understood that these operations or steps may not be performed in the order in which they appear in the embodiments of the present application or may be performed in parallel. The sequence numbers of the operations are only used to distinguish between different operations, and the sequence numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations or steps may be performed in sequence or in parallel, and these operations or steps may be combined.

[0023] First, some technical terms in this application are explained to facilitate those skilled in the art to understand this application.

[0024] Lattice-LSTM: Lattice structured- Long Short Term Memory, lattice structured long short-term memory network.

[0025] In order to make the objectives, technical solutions and advantages of this application clearer, the implementation methods of this application will be further described in detail below with reference to the accompanying drawings.

[0026] First, as Figure 1-Figure 2 As shown, the embodiment of the present application provides a method for constructing a knowledge graph of key technologies of intelligent mines, comprising the following steps: Step S1: Collect key common technical data on construction and operation from relevant databases in the field of smart mines to construct a data set; Step S2: Analyze and extract the characteristic information of the data set, and construct an ontology model of key common technologies for intelligent mine construction and operation containing a multi-level classification system; Step S3: Based on the constructed key common technology ontology model for intelligent mine construction and operation, a deep learning model integrating the BERT pre-trained language model and the Lattice-LSTM structure is used to perform entity recognition and relationship extraction on unstructured data to obtain entity relationship triples of entities, relationships, and attributes; Step S4: Visualize the entity relationship triple data and store it in a graph database to obtain a knowledge graph of key technologies for intelligent mines.

[0027] This application innovatively applies knowledge graph technology to the research on key technologies for smart mine construction and operation, and proposes a key technology knowledge graph construction strategy and application method suitable for the industry context. This can clarify the relevance, development path and innovation rules of technologies in the construction and operation of smart mines, and at the same time provide guidance and assistance to scholars and enterprises in related fields in understanding technology needs and development directions. By constructing an ontology of key technologies in intelligent mining and using the BERT-Lattice-LSTM model to extract named entities and entity relationships, the gap in the knowledge graph of key technologies in intelligent mining is filled. By identifying named entities in the field of mining technology and semi-automatically extracting unstructured knowledge based on the domain ontology, the visualization of entity interaction relationships and knowledge reasoning are realized.

[0028] In this application, for the convenience of description, the "Key Common Technology Knowledge Map for Smart Mine Construction and Operation" is referred to as the "Key Technology Knowledge Map for Smart Mine".

[0029] In one embodiment, step S1: collecting key common technical data on construction and operation from a database related to the smart mine field to construct a data set, specifically includes the following steps: Step S11: Constructing a smart mine technology knowledge graph database, including a specific database consisting of vertical field literature databases, patent databases, technical reports, and relevant national and local standards and specifications; Step S12: Preprocess the acquired technical knowledge graph database in the field of intelligent mining and construct a data set.

[0030] Based on step S1, by constructing a structured intelligent mining technology knowledge graph database, the systematic integration and standardized processing of multi-source heterogeneous data are achieved.

[0031] In one embodiment, the step S12: preprocessing the acquired technical knowledge graph database in the field of intelligent mining to construct a data set specifically includes the following steps: Step S121: Data cleaning is performed on the acquired technical knowledge graph database for the intelligent mining field. Noise information in the raw data is removed through a combination of regular expression matching, rule filtering, and manual verification. Specifically, this includes: removing non-text content such as HTML / XML tags, special characters, and garbled characters; standardizing formats such as dates and units (e.g., unifying "2023" to "2023"); correcting spelling errors (e.g., unifying "rock drill" with "rock drilling equipment"); and filtering low-quality data (e.g., deleting patent abstracts with fewer than 50 characters). For multi-source heterogeneous data (e.g., PDF reports and scanned documents), OCR recognition is used, followed by correction of recognition errors using the edit distance algorithm. Step S122: Perform Chinese word segmentation on the cleaned data and construct a domain-enhanced dictionary based on common word segmentation tools (such as Jieba and LTP). TF-IDF values ​​are used to extract high-frequency domain terms (such as "microseismic monitoring" and "digital twin"). Manual verification is performed based on the national standard "Mining Engineering Terminology" to ultimately form a domain dictionary of specialized terms. A bidirectional maximum matching algorithm is used to resolve ambiguities (e.g., "mining depth" takes precedence over "mining / depth"). Unknown terms are identified and supplemented using a BiLSTM-CRF model. Step S123: Perform BIO sequence annotation on the data after word segmentation processing to construct an annotated text set. Furthermore, for the relationship extraction task, an extended solution of BIO annotation is adopted to add relationship tags (such as "<mining process-applied to->intelligent working face") based on the entity annotation.

[0032] In one embodiment, step S2: analyzing and extracting feature information of a data set to construct an ontology model of key common technologies for intelligent mine construction and operation including a multi-level classification system, specifically includes the following steps: Step S21: extracting data features from the data set; Step S22: Establish a three-level core concept classification system of "key technologies - application scenarios - dependent facilities" based on the extracted data features; specifically, the key technology category includes 15 major categories of first-level maps, the application scenario category includes 5 major categories of first-level maps, and the dependent facilities category includes 2 major categories of first-level maps. Each major category of the first-level map is subdivided into multiple second-level maps. To clarify technology dependencies, the second-level map of the dependent facilities can be further subdivided into multiple third-level maps. Step S23: Based on the established three-level core concept classification system, a relationship system between concepts is constructed, and the attribute system and corresponding rules of each concept are defined to construct an ontology model of key common technologies for intelligent mine construction and operation containing a multi-level classification system; further, a domain core concept hierarchy, a hierarchy and attribute relationship between concepts are established, corresponding constraints are set, and protégé is used to fill the ontology model of key technologies for intelligent mines, and some class instances are added; specifically, in order to fully consider the intelligent application and dependent collaborative relationship of technology, this method refers to the "Intelligent Mine Data Fusion Specification" and related literature on intelligent mine system research, and uses Protégé software to map the key technology knowledge graph for intelligent mine construction. The spectrum is defined as three ontology categories: key technologies, application scenarios, and dependent facilities. The key technologies category encompasses 15 categories of digital technologies at the first level. The application management category includes five categories: safety assurance systems, data management systems, intelligent control systems, intelligent monitoring systems, and intelligent asset management. The dependent facilities category includes two first-level categories: basic equipment and infrastructure. These categories are further subdivided into multiple second- and third-level graphs. For example, basic equipment includes safety equipment, control and monitoring equipment, chute equipment, and transportation equipment. Control and monitoring equipment includes geological monitoring equipment, data acquisition equipment, and automated control equipment. When defining attribute and relationship constraints, object attributes, data attributes, and constraint rules are considered. For example, the entity "deep rock drilling robot" for "automated mining technology" has attribute definitions such as control precision = 0.1mm and max depth = 1500m. Constraint conflicts are also checked using Protégé's built-in inference engine (HermiT / Pellet). For example, if entities belonging to both "open-pit mining technology" and "underground mining technology" trigger a logical inconsistency alert.

[0033] In one embodiment, step S23: the specific composition of the ontology model of key common technologies for intelligent mine construction and operation includes concept class, attribute class, relationship class and rule class; wherein, the concept class includes three core categories of technology, system and facility, and each category has subdivided sub-concepts; the attribute class is used to describe the specific characteristic parameters of each concept, including technology maturity, system interface standards and facility technical indicators; the relationship class defines various association methods between concepts, including technical implementation relationships, system dependency relationships, etc.; the rule class includes various business constraints and logical reasoning rules, which together constitute a complete ontology model system.

[0034] In a more specific embodiment, the construction of the key common technology ontology of intelligent mine construction and operation in step S2 is as follows: Figure 3As shown, an ontology clearly defines the object entities, attributes, and relationships within a knowledge graph, clarifying its feasible scope. In this embodiment, the logical consistency and interpretability of the knowledge graph for key technologies in intelligent mining are enhanced by defining the technical concept hierarchy, attribute constraints, and semantic relationships. The core feature of an ontology is that it serves as a standardized expression of domain consensus. Through a conceptual hierarchy and semantic constraint rules, it constructs a shareable and reusable knowledge modeling template.

[0035] In this embodiment, the core value of the ontology is further focused on the construction requirements of the knowledge graph of key technologies in intelligent mines: (1) Clarify the hierarchical relationships and constraints among key technologies of smart mines (e.g., digital twins), technology application scenarios (e.g., intelligent control systems), and technology-dependent facilities (e.g., multi-sensor networks); (2) By constructing a “technology-scenario-facility” trinity relationship map, the knowledge redundancy in technical solution design can be reduced, for example, by avoiding repeated definitions of the semantics of “unmanned driving” in different scenarios; (3) Based on SWRL rules, the automatic derivation of technical knowledge links is realized, for example, reasoning from "rock stress data" to "support parameter optimization suggestions."

[0036] In one embodiment, step S3: based on the constructed key common technology ontology model for intelligent mine construction and operation, a deep learning model integrating a BERT pre-trained language model and a Lattice-LSTM structure is used to perform entity recognition and relationship extraction on unstructured data to obtain entity relationship triple data of entities, relationships, and attributes, specifically including the following steps: Step S31: Based on the constructed key common technology ontology model for intelligent mine construction and operation, the BERT encoding layer and Lattice-LSTM structure network are used to perform entity recognition and relationship extraction on the preprocessed data to obtain entity relationship triple data of entities, relationships and attributes; Step S32: extract triplet data through attention mechanism and CFR layer decoding optimization, and calculate the loss to train the ontology model until the model converges to obtain the optimized triplet data; Step S33: Perform entity disambiguation and entity fusion on the optimized triple data to generate triple data after entity disambiguation and entity fusion and store it in the knowledge base.

[0037] In one embodiment, in step S31, the specific training of the BERT-Lattice-LSTM model is implemented as follows: Initialize the parameters of the BERT module and Lattice-LSTM named entity recognition module of the BERT-Lattice-LSTM model, and initialize hyperparameters such as the learning rate and training rounds; The BERT-Lattice-LSTM model is trained and learned based on the training set, and the Adam optimization algorithm is used to optimize the parameters of the BERT model to avoid weight overload and improve the model's robustness in recognizing sparse entities (such as "microseismic monitoring signals").

[0038] This application proposes a Lattice-LSTM enhanced architecture that integrates a pre-trained language model, such as Figure 4 As shown in the figure, this model implements context-aware encoding of character-level features by introducing the BERT module and captures deep semantic associations through a multi-head attention mechanism. It also constructs a dual-channel input mechanism, where character features are encoded by BERT and then fed into LSTM units, while word-level features are distributedly represented using the Word2vec model. In the embodiment of extracting key technical knowledge in the field of intelligent mining, this method is implemented according to the following steps: (1) Implement multi-source heterogeneous data cleaning for electronic documents and network documents, including noise filtering, unstructured conversion and domain dictionary construction; (2) Use the BIOES annotation system for sequence annotation and combine it with the domain knowledge base for data enhancement. For example, [intelligent ventilation system [S]] is required in [deep mining [B] environment [I]]; (3) Initialize BERT parameters based on the transfer learning strategy and use domain corpus for adaptive fine-tuning to adapt to the contextual representation of smart mining domain terms. For high-frequency technical terms such as "digital twin" and "industrial Internet", increase their masking probability and strengthen the model's contextual reasoning ability for professional terms. (4) Based on character vectors and word vectors, the gating mechanism is used to weight the fusion of character and word sequence features to solve the boundary conflict problem of nested entities (such as "deep mining rock stress monitoring system" contains the sub-entity "stress monitoring"). At the same time, a label transfer matrix is ​​designed to prohibit illegal label sequences, such as "B-equipment→I-scene", so as to realize entity recognition and relationship extraction, and finally build a domain knowledge graph.

[0039] In one embodiment, the present application uses the BERT-Lattice-LSTM model to extract named entities and entity relationships from technical information in the field of smart mining. By fusing character vectors with word vectors, the boundary demarcation of professional vocabulary entities in this field is achieved. Specifically, the following steps are included: In order to achieve deep semantic understanding and accurate representation of smart mine technology texts, this step first uses texts in the field of smart mine technology as training corpus to fine-tune the BERT model for domain adaptability. This model can capture the contextualized semantic features of proprietary terms in the mining field through a bidirectional context encoding mechanism, such as the representation of complex concepts such as "rock stress monitoring" and "intelligent ventilation control." At the same time, a continuous bag-of-words architecture is used to train word-level vectors, predict target words through context windows, and strengthen the distributed semantic associations of domain vocabulary. For example, the vector space proximity modeling of related terms such as "rock drill" and "drilling efficiency" and "vibration monitoring" is performed; A Lattice-LSTM architecture was designed to dynamically fuse character and word sequence information to address the challenge of recognizing nested entities in mining text. For example, "stratum stress monitoring system" contains the subentity "stress monitoring." The parameters of the BERT module and Lattice-LSTM named entity recognition module of the BERT-Lattice-LSTM model were initialized, along with hyperparameters such as the learning rate and number of training rounds. The BERT-Lattice-LSTM model was trained and learned based on the training set. The Adam optimization algorithm was used to optimize the BERT model parameters to avoid weight overload and improve the model's robustness in recognizing sparse entities (such as "microseismic monitoring signal"). For the preprocessed smart mine technology text collection, a lattice structure was used to integrate the potential boundary information between character sequences and word sequences. Relationship extraction rules were designed based on the technical characteristics of smart mines, such as "technology-application scenario" and "equipment-functional parameters" correspondence rules, guiding the model to prioritize high-frequency association patterns. A bidirectional LSTM structure was used to model long-term dependencies within text sequences. For example, the forward LSTM captured the technical chain "deep mining → rock formation stress → real-time monitoring," while the backward LSTM analyzed the reverse logic of "ventilation system → energy consumption optimization → sensor network." A modified Lattice-LSTM model was used to identify semantic associations between entities based on an attention mechanism, enabling joint extraction of entity relationships.

[0040] In one embodiment, a standardized Cypher query statement is constructed based on the extracted "entity-relationship" triples to generate a CSV file of nodes and relationships. A Python script is used to batch convert the CSV file to UTF-8 encoding to avoid garbled Chinese characters. Automated node import is achieved through the LOAD CSV command, supporting dynamic label assignment and attribute mapping. Indexes are created for high-frequency relationships, such as "technology dependency" and "equipment collaboration," to accelerate query responses. D3.js force-directed simulation is used to define node repulsion (-300) and connection distance (150px), enabling data visualization and interactive functions for the knowledge graph of key common technologies for intelligent mine construction and operation.

[0041] In the vertical domain knowledge graph construction method, entity recognition and relationship extraction constitute the core processing links. To address the problem of blurred entity boundaries caused by the lack of explicit word segmentation tags in Chinese text, we first perform character-level segmentation processing on the original corpus to generate unlabeled character sequences; then we build a dual-channel feature fusion model based on deep learning. Its technical architecture includes three processing stages: (1) A bidirectional long short-term memory network (BiLSTM) is used to model the contextual semantics of character sequences and construct character-level distributed representations through embedding vector space mapping; (2) A temporal feature extraction module is constructed based on a multi-layer LSTM network, which captures long-range semantic dependencies through a gated loop mechanism; (3) Label inference optimization layer: Apply conditional random field (CRF) to establish the label transfer probability matrix, and obtain the optimal label sequence through the Viterbi decoding algorithm.

[0042] In one embodiment, step S4: visualizing the entity relationship triple data and storing it in a graph database to obtain a knowledge graph of key technologies for intelligent mines, specifically includes the following steps: Step S41: Convert the optimized triple data into graph database nodes and edges according to preset mapping rules, where key technology entities, application system entities, and dependent facility entities correspond to different types of nodes, and the relationships between the entities are converted into connecting edges; Step S42: Utilize the visualization function of the graph database to realize the intuitive display of the knowledge graph, and at the same time establish an efficient query channel through the database index mechanism to complete the structured storage and visualization presentation of key technical knowledge of intelligent mines.

[0043] In one embodiment, step S42 utilizes the visualization function of the graph database to realize the visualization display of the knowledge graph, which specifically includes the following steps: The concepts in the mapping ontology model are nodes, the mapping relationships are edges, and additional attributes are attached to the corresponding nodes. The D3.js visualization framework is used to visualize the data into a knowledge graph and store it in the Neo4j graph database. Among them, the Neo4j open database stores knowledge entities as nodes, uses node labels to represent the entity class to which they belong, and assigns multiple labels at the same time to represent knowledge entities of cross-cutting or integrated technologies and application types.

[0044] In a more specific embodiment, the data storage strategy of the knowledge graph of key technologies for intelligent mine construction based on Neo4j is specifically implemented as follows: In the knowledge graph, there are various representation and storage methods for large amounts of knowledge with high complexity, low structure, and strong connections. It is necessary to determine the appropriate storage and representation methods based on the characteristics of vertical domain knowledge and application scenarios. The graph database is a new type of NoSQL database based on graph model storage. The relationship storage structure and information query method between data are based on graph theory. As one of the many popular graph databases currently, Neo4j is very suitable for storing data with multiple interconnected relationships. Compared with traditional types of databases, Neo4j has the advantages of large storage capacity, good scalability, excellent portability, and detailed operation documents. Neo4j not only supports efficient and localized graph data storage, but also can solve the complex relationships between knowledge information, ensure the fast traversal of any graph algorithm, and avoid excessive resource consumption caused by the use of global indexes. Therefore, it is proposed to use Neo4j as the technical basis for the storage and query of the knowledge graph of key technologies for intelligent mine construction and operation. The key common technology knowledge graphs for intelligent mine construction and operation of some of this embodiment are as follows. Figure 5 shown.

[0045] Figure 5 (a) is part of the knowledge graph of key technologies in intelligent mining; Figure 5 (b) in the figure is an example diagram of the knowledge graph of dynamic simulation technology-intelligent control system.

[0046] The graph database Neo4j uses a labeled property graph model, storing knowledge entities as nodes, with node labels representing the entity class. Due to the non-unique nature of labels, multiple labels can be assigned simultaneously to represent cross-category knowledge entities. Table 1 shows the Neo4j storage strategy for named entities and relationships in the knowledge graph for key technologies in intelligent mine construction and operation.

[0047] Table 1 Knowledge storage strategy table

[0048] This application first completes the design of the knowledge graph architecture of key technologies for intelligent mines at the logical level, implements the classification of entity types within the field based on the similarities and differences of entity attributes, and defines constraints on the relationships between entity types at each level to ensure the standardized management of the knowledge base model; secondly, the Lattice-LSTM correction model integrated with BERT is used to perform named entity recognition and entity relationship extraction on the abstracts and network documents of literature patents in the field of intelligent mine construction and operation, and the extracted massive, multi-source key technical knowledge is stored in the Neo4j open database in a graph structure to complete the construction of the knowledge graph of key technologies for intelligent mines, and conduct intelligent application analysis and verification. The Lattice-LSTM correction model is an extension of the character-based BiLSTM-CRF model, integrating word-based cells and additional gates for controlling information flow, explicitly utilizing the relationship between word sequences, and integrating potential word information into the BiLSTM-CRF model, thereby improving the performance of named entity recognition.

[0049] This application studies the construction strategy of key technology knowledge graphs by combing through the architecture of intelligent mine construction systems. Combined with relevant literature and patents in the field of intelligent mine construction, demand research and analysis are carried out to clarify the daily use objects and group characteristics of the knowledge base, and design the storage strategy of entity concept classes, knowledge relationship classes and attributes in the knowledge base; analyze the knowledge characteristics of technologies related to intelligent mining, and explore the application of key technologies of knowledge graphs in the field of intelligent mining. Conduct a detailed analysis of the knowledge structure and characteristics of key technologies of intelligent mining, divide the types of technical named entities, and determine the named entities, entity attributes and relationship types; then use the extraction model to accurately extract knowledge entities and relationships, and realize large-scale knowledge storage of key technologies, application scenarios and dependent facilities in the field of intelligent mining based on the graph database; based on the research on knowledge graph construction strategies and key technologies, conduct analysis on the construction and application of intelligent mining technology knowledge graphs. Through experimental comparative research, verify the feasibility and practicality of the results, improve and optimize its shortcomings, and realize convenient query and visual management of knowledge in this field.

[0050] On the second aspect, the present application provides a knowledge graph construction system for key common technologies of intelligent mines, including a data acquisition and preprocessing module, an ontology model construction module, a knowledge extraction module and a knowledge storage and visualization module. The data acquisition and preprocessing module is used to collect key common technology data for construction and operation from relevant databases in the field of intelligent mines and construct a data set; the ontology model construction module is used to analyze and extract feature information of the data set and construct an ontology model of key common technologies for construction and operation of intelligent mines containing a multi-level classification system; the knowledge extraction module is used to perform entity recognition and relationship extraction on unstructured data based on the constructed ontology model of key common technologies for construction and operation of intelligent mines, using a deep learning model that integrates the BERT pre-trained language model and the Lattice-LSTM structure to obtain entity relationship triplet data of entities, relationships and attributes; the knowledge storage and visualization module is used to visualize the entity relationship triplet data and store it in a graph database to obtain a knowledge graph of key technologies for intelligent mines.

[0051] Furthermore, the ontology model building module includes: A feature extraction unit, used to extract data features from a data set; A concept classification system establishment unit, in communication with the feature extraction unit, for establishing a three-level core concept classification system of "key technology - application scenario - dependent facilities" based on the extracted data features; The ontology model construction unit, based on the established three-level core concept classification system, constructs an association system between concepts, and defines the attribute system and corresponding rules of each concept, thereby constructing an ontology model of key common technologies for intelligent mine construction and operation that includes a multi-level classification system.

[0052] In a third aspect, the present application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the method for constructing a knowledge graph of key common technologies for intelligent mines as described above.

[0053] Among them, the functional implementation of each module in the above-mentioned intelligent mine key technology knowledge graph construction system corresponds to the various steps in the above-mentioned intelligent mine key technology knowledge graph construction method embodiment, and its functions and implementation process will not be repeated here one by one.

[0054] On the third aspect, an embodiment of the present application provides a device for constructing a knowledge graph of key technologies for intelligent mines. The device for constructing a knowledge graph of key technologies for intelligent mines can be a personal computer (PC), a laptop, a server, or other device with data processing capabilities.

[0055] Communication interfaces include input / output (I / O), physical, and logical interfaces, which interconnect components within the equipment used to build the knowledge graph for key technologies in intelligent mining, as well as interfaces that connect the equipment to other devices (such as other computing devices or user devices). Physical interfaces can include Ethernet, fiber, and ATM interfaces; user devices can include displays and keyboards.

[0056] The memory can be various types of storage media, such as random access memory (RAM), read-only memory (ROM), non-volatile RAM (NVRAM), flash memory, optical storage, hard disk, programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), etc.

[0057] The processor may be a general-purpose processor that can invoke a program for constructing a knowledge graph for key technologies in intelligent mining stored in a memory and execute the method for constructing a knowledge graph for key technologies in intelligent mining provided in the embodiments of this application. For example, the general-purpose processor may be a central processing unit (CPU). The method executed when the program for constructing a knowledge graph for key technologies in intelligent mining is invoked can be referenced in the various embodiments of the method for constructing a knowledge graph for key technologies in intelligent mining provided in this application and will not be further described here.

[0058] In a fourth aspect, an embodiment of the present application also provides a readable storage medium.

[0059] The readable storage medium of the present application stores a program for constructing a knowledge graph of key technologies for intelligent mines, wherein when the program for constructing a knowledge graph of key technologies for intelligent mines is executed by a processor, the steps of the method for constructing a knowledge graph of key technologies for intelligent mines as described above are implemented.

[0060] Among them, the method implemented when the construction program of the key technology knowledge graph of intelligent mines is executed can refer to the various embodiments of the construction method of the key technology knowledge graph of intelligent mines in this application, and will not be repeated here.

[0061] It should be noted that the serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0062] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, or the part that contributes to the existing technology, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above and includes a number of instructions for enabling a terminal device to execute the methods described in each embodiment of this application.

[0063] The above are only preferred embodiments of the present application and do not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for constructing a knowledge graph of key common technologies for intelligent mines, characterized by: The following steps are involved: Collect key common technical data on construction and operation from relevant databases in the field of smart mines and build a data set; Analyze and extract characteristic information from data sets, and construct an ontology model of key common technologies for intelligent mine construction and operation, including a multi-level classification system; Based on the constructed ontology model of key common technologies for intelligent mine construction and operation, a deep learning model integrating the BERT pre-trained language model and the Lattice-LSTM structure is used to perform entity recognition and relationship extraction on unstructured data, obtaining entity relationship triples of entities, relationships, and attributes. The entity relationship triplet data is visualized and stored in a graph database to obtain a knowledge graph of key technologies in intelligent mining.

2. The method for constructing a knowledge graph of key common technologies for intelligent mines according to claim 1, characterized in that: The process of collecting key common technical data on construction and operation from relevant databases in the field of intelligent mining and constructing a data set specifically includes the following steps: Build a knowledge graph database for intelligent mining technology; Preprocess the data source of knowledge related to key technologies of intelligent mines and construct a data set.

3. The method for constructing a knowledge graph of key common technologies for intelligent mines according to claim 1, characterized in that: The analysis extracts characteristic information of the data set and constructs a key common technology ontology model for intelligent mine construction and operation containing a multi-level classification system, specifically including the following steps: Extract data features from the dataset; Establish a three-level core concept classification system of "key technologies - application scenarios - dependent facilities" based on the extracted data features; Based on the established three-level core concept classification system, a relationship system between concepts is constructed, and the attribute system and corresponding rules of each concept are defined to construct a key common technology ontology model for intelligent mine construction and operation that includes a multi-level classification system.

4. The method for constructing a knowledge graph of key common technologies for intelligent mines according to claim 3, characterized in that: The specific composition of the ontology model of key common technologies for intelligent mine construction and operation includes concept class, attribute class, relationship class and rule class; among them, the concept class includes three core categories: key technologies, application scenarios and dependent facilities, and each category has sub-concepts; the attribute class is used to describe the specific characteristic parameters of each concept, including technical methods, system optimization goals and facility deployment costs; the relationship class defines various association methods between concepts, including technical dependency relationships, equipment collaborative relationships, etc.; the rule class includes the constraints of various entities and the relationships between entities, as well as logical reasoning rules, which together constitute a complete ontology model system.

5. The method for constructing a knowledge graph of key common technologies for intelligent mines according to claim 1, characterized in that: The method, based on the key common technology ontology model for intelligent mine construction and operation, uses a deep learning model integrating the BERT pre-trained language model and the Lattice-LSTM structure to extract knowledge from unstructured data and obtain entity relationship triple data of entities, relationships, and attributes. Specifically, the method includes the following steps: Based on the constructed ontology model of key common technologies for intelligent mine construction and operation, the BERT encoding layer and Lattice-LSTM structure network are used to perform entity recognition and relationship extraction on the preprocessed data to obtain entity relationship triple data of entities, relationships and attributes; Extract triplet data through attention mechanism and CFR layer decoding optimization, and calculate loss to train the ontology model until the model converges to obtain optimized triplet data; Entity disambiguation and entity fusion are performed on the optimized triple data to generate triple data after entity disambiguation and entity fusion.

6. The method for constructing a knowledge graph of key common technologies for intelligent mines according to claim 1, characterized in that: The visualization of the entity relationship triple data and storing it in a graph database to obtain a knowledge graph of key technologies of intelligent mines specifically includes the following steps: The optimized triple data is converted into graph database nodes and edges according to the preset mapping rules. Key technology entities, application system entities, and dependent facility entities correspond to different types of nodes, and the relationships between these entities are converted into connecting edges. The visualization function of the graph database is used to achieve an intuitive display of the knowledge graph. At the same time, an efficient query channel is established through the database indexing mechanism to complete the structured storage and visual presentation of key technical knowledge of intelligent mines.

7. The method for constructing a knowledge graph of key common technologies for intelligent mines according to claim 6, characterized in that: The visualization function of the graph database is used to realize the visualization of the knowledge graph, which specifically includes the following steps: The concepts in the mapping ontology model are nodes, the mapping relationships are edges, and additional attributes are added to the corresponding nodes. The D3.js visualization framework is used to visualize the data into a knowledge graph and store it in the Neo4j graph database.

8. A system for constructing a knowledge graph of key common technologies for intelligent mines, characterized by: include: The data acquisition and preprocessing module is used to collect key common technical data on construction and operation from relevant databases in the field of intelligent mining and construct a data set; An ontology model building module is used to analyze and extract characteristic information from data sets and construct an ontology model of key common technologies for intelligent mine construction and operation, which includes a multi-level classification system. The knowledge extraction module is used to perform entity recognition and relationship extraction on unstructured data based on the constructed key common technology ontology model for intelligent mine construction and operation, using a deep learning model that integrates the BERT pre-trained language model and the Lattice-LSTM structure to obtain entity-relationship triple data of entities, relationships, and attributes; The knowledge storage and visualization module is used to visualize the entity relationship triple data and store it in a graph database to obtain a knowledge graph of key technologies in intelligent mines.

9. The system for constructing a knowledge graph of key common technologies for intelligent mines according to claim 8, characterized in that: The ontology model building module includes: A feature extraction unit, used to extract data features from a data set; A concept classification system establishment unit, in communication with the feature extraction unit, for establishing a three-level core concept classification system of "key technology - application scenario - dependent facilities" based on the extracted data features; The ontology model construction unit, based on the three-level core concept classification system, constructs an association system between concepts, defines the attribute system of each concept and the corresponding rules, and constructs an ontology model of key common technologies for intelligent mine construction and operation that includes a multi-level classification system.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for constructing a knowledge graph of key common technologies for intelligent mines as described in any one of claims 1 to 7.

Images