Metallization prediction model construction method and system based on knowledge graph, and product

By constructing a mineralization prediction model based on knowledge graphs, the correlation and implicit connections of mineral resource-related data are explored, which solves the problem of the lack of universality and generality of existing models and realizes the improvement of the intelligence and accuracy of mineral resource prediction.

CN119004247AActive Publication Date: 2024-11-22CHINA UNIV OF GEOSCIENCES (BEIJING)

Patent Information

Application Number
CN202410886042.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-03
Publication Date
2024-11-22
Estimated Expiration
2044-07-03

AI Technical Summary

Technical Problem

Existing mineral resource prediction models lack comprehensive utilization of factors such as lithospheric material structure, crust-mantle evolution, and ore-forming material evolution and cycle, resulting in prediction results that lack universality and versatility and are difficult to meet the requirements of intelligent prediction.

Method used

We construct a mineralization prediction model based on knowledge graphs. By collecting and labeling mineral resource-related data, we use machine learning technology to build a data generation model, mine the relationships and implicit connections between entities, and construct a full-factor prediction model.

Benefits of technology

It has enabled the comprehensive utilization of mineral resource elements, significantly improved the accuracy of forecast results, and met the intelligent requirements of mineral resource forecasting.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004247A_ABST
    Figure CN119004247A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of mineral resource prediction, and discloses a mapping knowledge domain-based mineralization prediction model construction method and system and a product, and the method comprises the steps: constructing a mapping knowledge domain of mineral resources, the mapping knowledge domain comprising knowledge data of elements, minerals, rocks, ore deposits, mineralization and geological phenomena; performing knowledge extraction on the knowledge graph based on the data types of the earth system, the mineralization system, the exploration system and the prediction evaluation system to obtain first knowledge data; performing knowledge mining on the first knowledge data and / or the knowledge graph based on a preset knowledge mining method to obtain second knowledge data; and constructing a mineral resource prediction model based on the first knowledge data and the second knowledge data. According to the method, the influence of the lithosphere material structure, the shell curtain evolution, the mineral substance evolution and circulation and other factors on mineralization can be considered, the comprehensive utilization of mineral resource prediction elements is realized, the accuracy of the prediction result is improved, and the intelligent requirement of mineral resource prediction is met.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of mineral resource prediction, and in particular to a method, system and product for constructing a mineralization prediction model based on a knowledge graph. Background Art

[0002] Mineral resources refer to the collection of minerals or useful elements that are naturally present in the crust or on the surface of the earth and have development and utilization value, which are formed through geological mineralization. Mineral resources are an important foundation and valuable window for promoting the development of human society and exploring the operation mode of the earth system. It is of great significance to accurately predict them.

[0003] At present, relevant mineral resource prediction research is mainly based on expert experience and established typical mineral deposit models, using geological, geophysical, geochemical, remote sensing data, etc. to estimate the probability of mineralization in the target area. In this process, such research often only focuses on important indicators directly related to mineralization.

[0004] However, the mineralization system is not an isolated system, but a coupled product of the interconnected, mutually influential and mutually constrained physical, chemical and biological effects of the Earth's multiple spheres. Existing mineral resource prediction models pay little attention to the remote dynamic correlation and mutual constraints of related factors such as the lithospheric material structure, crust-mantle evolution, and the evolution and circulation of mineralizing materials. They are unable to fully utilize mineral resource elements for effective prediction, resulting in the establishment of mineralization models and prediction models that are mostly exclusive and special, lacking universality and versatility. Summary of the invention

[0005] In view of this, the present invention provides a method, system and product for constructing a mineralization prediction model based on a knowledge graph to solve the problem that the existing mineral resource prediction model construction lacks comprehensive utilization of mineral resource elements, affects the accuracy of the prediction results, and is difficult to meet the requirements of intelligent mineral resource prediction.

[0006] In a first aspect, the present invention provides a method for constructing a mineralization prediction model based on a knowledge graph, the method comprising:

[0007] Construct a knowledge graph of mineral resources, which contains knowledge data on elements, minerals, rocks, ore deposits, mineralization and geological phenomena;

[0008] Based on the data types of the earth system, the mineralization system, the exploration system, and the prediction and evaluation system, knowledge is extracted from the knowledge graph to obtain the first knowledge data;

[0009] Performing knowledge mining on the first knowledge data and / or the knowledge graph based on a preset knowledge mining method to obtain second knowledge data;

[0010] A mineral resource prediction model is constructed based on the first knowledge data and the second knowledge data.

[0011] The present invention takes into account the influence of factors such as lithospheric material structure, crust-mantle evolution, evolution and circulation of ore-forming materials on mineralization, obtains corresponding knowledge data based on earth system data type, mineralization system data type, exploration system data type and prediction and evaluation system data type, and further mines new knowledge data from the data to construct a mineral resource prediction model, thereby realizing the comprehensive utilization of mineral resource elements, significantly improving the accuracy of prediction results, and meeting the intelligent requirements of mineral resource prediction.

[0012] In an optional implementation, constructing a knowledge graph of mineral resources includes:

[0013] Collect relevant data on mineral resources including earth system, mineralization system, exploration system and prediction and evaluation system;

[0014] The relevant data are annotated with entities and relationships to obtain a large amount of knowledge data. The knowledge data is triple data containing entities and entity relationships. Entities represent elements related to mineralization of mineral resources, and entity relationships represent semantic relationships between different entities.

[0015] Extracting training data and test data from a large amount of knowledge data, and building a data generation model based on the training data, wherein the data generation model is used to generate triple data containing entities and entity relationships;

[0016] The test data is input into the data generation model for prediction, correspondingly obtaining a large amount of generated triple data;

[0017] All triple data are stored to obtain the knowledge graph of mineral resources.

[0018] The present invention collects mineral resource-related data and labels it to obtain standard triple knowledge data, and predicts the labeled knowledge data through a constructed machine learning model to generate massive triple data. This can ensure the standardization and richness of the data set and improve the accuracy of the prediction results of the mineral resource prediction model to a certain extent.

[0019] In an optional implementation, the second knowledge data includes the first mined knowledge, and knowledge mining is performed on the first knowledge data and / or the knowledge graph based on a preset knowledge mining method to obtain the second knowledge data, including:

[0020] In response to two target entities selected by the user from the first knowledge data and / or the knowledge graph, knowledge search is performed on the two target entities based on a preset entity relationship type query method to obtain corresponding search knowledge;

[0021] Integrate the search knowledge to obtain the first mining knowledge.

[0022] When acquiring a target entity given by a user, the present invention uses knowledge retrieval technology to quickly find the same or most similar knowledge data from the relevant knowledge graph, further mines the strong and indirect entity relationships between entities, and identifies the long-range and implicitly connected entity relationships between entities, providing data indications and reference content for the subsequent construction of a full-factor mineral resource prediction model.

[0023] In an optional implementation, the preset entity relationship types include direct relationships and indirect relationships, wherein the indirect relationships include short-range relationships and long-range relationships; performing knowledge search on two target entities based on the preset entity relationship type query method, and obtaining corresponding search knowledge includes:

[0024] Get the minimum number of hops and paths required for the two target entities to be associated;

[0025] Determine whether the minimum hop count is less than a preset hop count threshold;

[0026] If the minimum hop count is less than a preset hop count threshold, it is determined that there is a direct relationship between the two target entities, and corresponding search knowledge is obtained based on the two target entities and the direct relationship;

[0027] If the minimum hop count is not less than the preset hop count threshold, it is determined that the two target entities have an indirect relationship, and it is determined whether the number of paths is less than the preset path threshold;

[0028] If the number of paths is less than a preset path threshold, it is determined that there is a short-range relationship between the two target entities, and corresponding search knowledge is obtained based on the two target entities and the short-range relationship;

[0029] If the number of paths is not less than a preset path threshold, it is determined that there is a long-range relationship between the two target entities, and corresponding search knowledge is obtained based on the two target entities and the long-range relationship.

[0030] The present invention quantitatively determines the association relationship between entities through the minimum number of hops and the number of paths between entities, can mine richer entity relationships, and provide research data for the construction of subsequent full-factor mineral resource prediction models.

[0031] In an optional implementation, the second knowledge data includes second mined knowledge, and the first knowledge data and / or the knowledge graph are mined based on a preset knowledge mining method to obtain the second knowledge data, including:

[0032] Constructing a data set based on the first knowledge data and / or the knowledge graph, the data set comprising: a training data set and a test data set;

[0033] The training data set is trained through a preset knowledge graph embedding model to obtain a corresponding knowledge reasoning model;

[0034] Input the test data set into the knowledge reasoning model for reasoning to obtain reasoning knowledge;

[0035] Integrate the reasoning knowledge to obtain the second mining knowledge.

[0036] The present invention is based on existing knowledge data and uses the knowledge graph embedding model to perform knowledge reasoning on it, which can mine new possible entity relationships, namely reasoning knowledge, and provide research data for the subsequent construction of a full-factor mineral resource prediction model.

[0037] In an optional implementation, the second knowledge data includes third mining knowledge, and the first knowledge data and / or knowledge graph are mined based on a preset knowledge mining method to obtain the second knowledge data, including:

[0038] Obtain all paths where the mineral resources are associated with any tail entity in the first knowledge data and / or the knowledge graph;

[0039] Count the occurrence frequency of different entity relationships in each path, calculate the relationship probability between mineral resources and tail entities, and determine the intensity index of the corresponding path based on the occurrence frequency and relationship probability;

[0040] The strength indexes of all paths are integrated to obtain the third mining knowledge.

[0041] The present invention takes into account the different probabilities of mineralization influence of different elements, minerals and rocks on mineral resources, determines the intensity index of the corresponding path based on the frequency of occurrence and relationship probability of the association path between entities, and characterizes the strength of the association relationship between corresponding entities through the intensity index. It can realize the calculation of new implicit knowledge based on graph analysis and perform knowledge mining with different weight assignments, thereby providing research data for the subsequent construction of a full-factor mineral resource prediction model.

[0042] In a second aspect, the present invention provides a system for constructing a mineralization prediction model based on a knowledge graph, the system comprising:

[0043] Data construction module, used to construct the knowledge graph of mineral resources, which contains knowledge data of elements, minerals, rocks, ore deposits, mineralization and geological phenomena;

[0044] A knowledge extraction module is used to extract knowledge from the knowledge graph based on the data types of the earth system, the mineralization system, the exploration system, and the prediction and evaluation system to obtain first knowledge data;

[0045] A knowledge mining module, configured to perform knowledge mining on the first knowledge data and / or the knowledge graph based on a preset knowledge mining method to obtain second knowledge data;

[0046] The model building module is used to build a mineral resource prediction model based on the first knowledge data and the second knowledge data.

[0047] The knowledge graph-based mineralization prediction model construction system of the present invention takes into account the influence of factors such as the lithospheric material structure, crust-mantle evolution, and the evolution and circulation of ore-forming materials on mineralization, and obtains corresponding knowledge data based on the earth system data type, mineralization system data type, exploration system data type, and prediction and evaluation system data type, and further mines new knowledge data from the data to construct a mineral resource prediction model, thereby realizing the comprehensive utilization of mineral resource elements, helping to improve the accuracy of prediction results, and meeting the intelligent requirements of mineral resource prediction.

[0048] In the third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to execute a method for constructing a mineralization prediction model based on a knowledge graph according to the first aspect or any corresponding embodiment thereof.

[0049] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute a method for constructing a mineralization prediction model based on a knowledge graph according to the first aspect or any corresponding embodiment thereof.

[0050] In a fifth aspect, the present invention provides a computer program product, comprising computer instructions, which are used to enable a computer to execute a method for constructing a mineralization prediction model based on a knowledge graph according to the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0052] Figure 1 It is a schematic diagram of the process of constructing a mineralization prediction model based on a knowledge graph according to an embodiment of the present invention;

[0053] Figure 2 It is a schematic flow chart of another method for constructing a mineralization prediction model based on a knowledge graph according to an embodiment of the present invention;

[0054] Figure 3 It is a schematic diagram of the results of knowledge graph query;

[0055] Figure 4 It is a schematic diagram of the result of knowledge reasoning;

[0056] Figure 5 It is a schematic diagram of calculation based on graph analysis;

[0057] Figure 6 It is a schematic diagram of the structure of the full-factor prediction model for porphyry copper deposits;

[0058] Figure 7 It is a structural block diagram of a system for building a mineralization prediction model based on a knowledge graph according to an embodiment of the present invention;

[0059] Figure 8 It is a schematic diagram of the structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0061] Mineral resources are an important foundation and a valuable window for promoting the development of human society and exploring the operating mode of the earth system. However, current mineral resource prediction research is mainly based on expert experience and established typical ore deposit models, using geological, geophysical, geochemical, remote sensing data, etc. to estimate the probability of mineralization in the target area. In this process, such research only focuses on important indicators directly related to mineralization. Since the mineralization system is a coupled product of the interconnection, mutual influence, and mutual restriction of the physical, chemical, biological and other processes of the earth's subsystems, it involves a large number of physical elements and corresponding physical relationships. The aforementioned mineral resource prediction model pays less attention to the remote dynamic correlation and mutual restriction of related factors such as the lithospheric material structure, crust and mantle evolution, and the evolution and circulation of mineralizing materials, resulting in the establishment of mineralization models and prediction models that are mostly exclusive and special, lacking universality and versatility.

[0062] In addition, with the explosive growth of geoscience big data and the rapid development of data science, it has also brought opportunities for the paradigm shift of earth science, and large-scale knowledge graph technology has been widely used in various industries. Knowledge reasoning and knowledge discovery based on knowledge graphs, as well as data-driven artificial intelligence technologies represented by machine learning, are reactivating and enhancing the ability of earth science to solve new problems. However, the application of mineral resource knowledge graphs in the past has remained at the level of simple knowledge query and inductive summary, and the constructed knowledge graphs are not rich and have one-sided elements. Therefore, an intelligent construction system solution for the full-factor prediction index system of mineral resources based on knowledge graph technology is expected.

[0063] Therefore, the present invention proposes a method, system and product for constructing a mineralization prediction model based on a knowledge graph, uses a knowledge graph to break through the complex mineralization system research, constructs the dynamic association and mutual constraint between the mineralization system and all mineralization-related factors, and uses the knowledge graph data-driven technology to realize the expansion of "Earth system-metallogenic system-exploration system-prediction and evaluation system", from "static correlation analysis of key elements" to "dynamic knowledge reasoning and discovery of all elements across scales". The present invention constructs a knowledge graph that expresses the complex earth system related to mineralization, associates deep geological processes-magma evolution-element migration and enrichment-ore mineral accumulation and other mechanisms, and uses machine learning and other technologies to build an inference engine to reveal the hidden factors and long-range factors related to porphyry mineralization in the earth system, which can realize the comprehensive utilization of mineral resource elements, significantly improve the accuracy of the prediction results, and meet the intelligent requirements of mineral resource prediction to a great extent.

[0064] An embodiment of the present invention provides an embodiment of a method for constructing a mineralization prediction model based on a knowledge graph. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0065] In this embodiment, a method for constructing a mineralization prediction model based on a knowledge graph is provided. Figure 1 is a flow chart of a method for constructing a mineralization prediction model based on a knowledge graph according to an embodiment of the present invention, such as Figure 1 As shown, the process includes the following steps:

[0066] Step S101, constructing a knowledge graph of mineral resources, which contains knowledge data on elements, minerals, rocks, ore deposits, mineralization and geological phenomena.

[0067] It should be noted that the specific types of mineral resources are not limited here, and can be adjusted according to actual needs. For example, the mineral resources are porphyry copper ore, which is only used as an example and is not limited thereto.

[0068] In a specific embodiment, the knowledge graph includes 12 types of knowledge data, including elements, minerals, rocks, hydrothermal alteration, ore deposits, structures, geophysics, geochemistry, geochronology, deep geological processes, exploration and evaluation data, and prediction models and methods. The specific content of each type of knowledge data can be determined based on the adaptability of actual project needs. For example, the inclusion of iron elements is only for illustrative purposes.

[0069] Step S102, extracting knowledge from the knowledge graph based on the data types of the earth system, mineralization system, exploration system and prediction and evaluation system to obtain first knowledge data.

[0070] It should be noted that the Earth system describes the connection between things from the macroscopic to the microscopic. It emphasizes starting from the whole. For example, the atmosphere, hydrosphere, lithosphere and biosphere are regarded as an organically connected Earth system. The global changes of various time scales occurring in this system are the result of the interaction between the various spheres of the Earth system and the interaction between humans and the environment (life and non-life). In-depth research on it will fundamentally understand the formation and evolution of the Earth and its future development trend, so as to better serve the rational use of resources and the improvement of the global environment for mankind, and strive to achieve the harmonious development of man and nature. The metallogenic system refers to the whole of all geological elements and metallogenic dynamic processes that control the formation and preservation of ore deposits in a certain time-space domain, as well as the formed ore deposit series and anomaly series. It is a natural system with metallogenic functions. The purpose of metallogenic system research is to understand the laws of metallogenesis and better guide mineral exploration. The exploration system is a series of work related to mineral exploration. It is guided by geological metallogenic theory, based on existing geological, mineral, geophysical, geochemical, remote sensing and other data and information, and uses mineral exploration and GIS technology to predict, evaluate and select target areas for mineral resources, and then implement drilling, pit exploration and other projects to discover mineral deposits and find out their quantity, quality and development and utilization conditions, so as to meet all geological exploration work required by actual construction. The prediction and evaluation system is a related evaluation system of the above system, such as the evaluation of geological exploration work.

[0071] In this embodiment, based on expert knowledge, knowledge can be extracted from the data types of four aspects: earth system, mineralization system, exploration system, and prediction and evaluation system to obtain corresponding knowledge data.

[0072] Step S103: perform knowledge mining on the first knowledge data and / or knowledge graph based on a preset knowledge mining method to obtain second knowledge data.

[0073] It should be noted that the specific content of the preset knowledge mining method in this embodiment is not described in detail here, and is adaptively selected based on actual project requirements and conventional operating means in the field. For example, the preset knowledge mining method includes a knowledge graph search method, a knowledge reasoning method, a knowledge completion method, a knowledge graph analysis method, etc., which are only used as an exemplary description.

[0074] Step S104: constructing a mineral resource prediction model based on the first knowledge data and the second knowledge data.

[0075] It should be noted that, in this embodiment, after the mineral resource prediction model is constructed, the target mineral resources can be predicted based on actual project requirements.

[0076] In summary, the method for constructing a mineralization prediction model based on a knowledge graph in an embodiment of the present invention takes into account the influence of factors such as the lithospheric material structure, crust-mantle evolution, and the evolution and circulation of mineralizing materials on mineralization, and obtains corresponding knowledge data based on the earth system data type, mineralizing system data type, exploration system data type, and prediction and evaluation system data type, and further mines new knowledge data from the data to construct a mineral resource prediction model, thereby realizing the comprehensive utilization of mineral resource elements, significantly improving the accuracy of prediction results, and meeting the intelligent requirements of mineral resource prediction.

[0077] In this embodiment, a method for constructing a mineralization prediction model based on a knowledge graph is provided. Figure 2 is a flow chart of another method for constructing a mineralization prediction model based on a knowledge graph according to an embodiment of the present invention, such as Figure 2 As shown, the process includes the following steps:

[0078] Step S201, constructing a knowledge graph of mineral resources, the knowledge graph contains knowledge data of elements, minerals, rocks, ore deposits, mineralization and geological phenomena.

[0079] Specifically, the above step S201 includes:

[0080] Step S2011, collect relevant data on mineral resources including earth system, mineralization system, exploration system and prediction and evaluation system.

[0081] In this embodiment, the specific content and collection methods of mineral resource data are not limited here, and are determined based on actual project requirements. For example, the mineral resources of porphyry copper mines can be collected and collated from public porphyry copper mine-related research literature or geological survey data, which is only used as an example.

[0082] It should be noted that the types of mineral resource data include text, images, tables and other types, which need to be converted into a unified knowledge data format, namely triple data.

[0083] Step S2012, annotate the entities and relationships of the relevant data to obtain a large amount of knowledge data, which is triple data containing entities and entity relationships, wherein entities represent elements related to mineralization of mineral resources, and entity relationships represent semantic relationships between different entities.

[0084] It should be noted that entities are one of the basic components of knowledge graphs, and they represent objects or concepts in the real world. Entities can be specific people, places, organizations, events, or abstract concepts, attributes, or categories. Each entity has a unique identifier for accurate identification and reference in the knowledge graph. In the knowledge graph, entities are usually represented as nodes and connected to other entities through edges, which represent the relationship between entities. Triples are a method of describing directed relationships between entities, also known as relationship triples, which are generally expressed as: <head entity, relationship, tail entity>.

[0085] Step S2013, extracting training data and test data from a large amount of knowledge data, and constructing a data generation model based on the training data, wherein the data generation model is used to generate triple data including entities and entity relationships.

[0086] In this embodiment, the triple data is predicted by building a machine learning model (i.e., a trained data generation model) to achieve the expansion of sample data. Specifically, the specific type of machine learning model is not limited here, and it is adaptively adjusted based on actual needs, such as a convolutional neural network model, which is only used as an exemplary illustration.

[0087] Step S2014, input the test data into the data generation model for prediction, and correspondingly obtain a large amount of generated triple data.

[0088] Step S2015, store all triple data to obtain the knowledge graph of mineral resources.

[0089] In a specific embodiment, the knowledge graph of mineral resources with porphyry copper as the core may be composed of millions of triple data, where each triple data contains two entities and one relationship; different entity types and relationship types are defined based on expert knowledge and theoretical models, and each entity type may contain hundreds to thousands of entities.

[0090] In the embodiment of the present invention, by collecting mineral resource related data and annotating it to obtain standard triple knowledge data, and predicting the annotated knowledge data through a constructed machine learning model to generate massive triple data, the standardization and richness of the data set can be guaranteed, and the accuracy of the prediction results of the mineral resource prediction model is improved to a certain extent.

[0091] Step S202: extract knowledge from the knowledge graph based on the data types of the earth system, mineralization system, exploration system, and prediction and evaluation system to obtain first knowledge data. Figure 1 Step S102 of the illustrated embodiment will not be described in detail here.

[0092] Step S203: perform knowledge mining on the first knowledge data and / or knowledge graph based on a preset knowledge mining method to obtain second knowledge data.

[0093] In this embodiment, the second knowledge data includes the first mining knowledge, the second mining knowledge and the third mining knowledge, and the aforementioned knowledge data are all obtained through the corresponding knowledge graph technology. Specifically, the three knowledge mining methods based on the knowledge graph in this embodiment include: (1) long-range and implicit associations based on knowledge graph search; (2) knowledge reasoning and knowledge completion based on the knowledge graph embedding model; (3) implicit knowledge discovery based on graph analysis calculation, which is only used as an exemplary illustration.

[0094] Specifically, when the second knowledge data is the first mined knowledge, the above step S203 includes:

[0095] Step A1, in response to two target entities selected by the user from the first knowledge data and / or the knowledge graph, knowledge search is performed on the two target entities based on a preset entity relationship type query method to obtain corresponding search knowledge.

[0096] It should be noted that the method for users to select target entities in this embodiment is not limited here and can be adaptively adjusted based on actual needs. For example, a corresponding query interface is developed based on the currently stored knowledge graph data to implement various query methods such as indirect, fuzzy, and composite for different entities, which is only used as an example.

[0097] In this embodiment, the preset entity relationship types include direct relationships and indirect relationships, wherein indirect relationships include short-range relationships and long-range relationships. It should be noted that if two entities can be associated with each other in only one hop (i.e., the number of hops), it is considered that there may be a direct relationship between them; if the minimum number of hops required for the association of two entities is greater than a set threshold, it is considered that there may be an indirect relationship between them. In addition, two entities may be linked through other different entities, that is, there are multiple association paths. As the number of hops between the two entities increases, their relationship changes from direct to indirect, and the association path changes from short-range to long-range; at the same time, with the increase in the number of paths, the relationship between the two entities will become increasingly complex from simple.

[0098] Specifically, in the above step A1, knowledge search is performed on the two target entities based on a preset entity relationship type query method, and corresponding search knowledge is obtained, including:

[0099] Step A11, obtaining the minimum number of hops and paths required for the association between two target entities.

[0100] Step A12, determining whether the minimum hop count is less than a preset hop count threshold.

[0101] In this embodiment, the specific value of the preset hop count threshold is not limited here, and is adaptively selected based on actual needs. For example, the preset hop count threshold is 2, which is only used as an exemplary description.

[0102] Step A13: If the minimum hop count is less than a preset hop count threshold, it is determined that there is a direct relationship between the two target entities, and corresponding search knowledge is obtained based on the two target entities and the direct relationship.

[0103] It should be noted that the search knowledge in this embodiment is also binary data including different entities and their corresponding entity relationships.

[0104] Step A14: if the minimum hop count is not less than the preset hop count threshold, it is determined that there is an indirect relationship between the two target entities, and it is determined whether the number of paths is less than the preset path threshold.

[0105] In this embodiment, when two target entities are indirectly related, there may be multiple association paths from the head target entity to the tail target entity, so it is necessary to further divide the entity relationship based on the number of association paths. It should be noted that the specific value of the preset path threshold is not limited here and is adaptively selected based on actual needs. For example, the preset path threshold is 3, which is only used as an exemplary description.

[0106] Step A15: If the number of paths is less than the preset path threshold, it is determined that there is a short-range relationship between the two target entities, and corresponding search knowledge is obtained based on the two target entities and the short-range relationship.

[0107] Step A16: If the number of paths is not less than the preset path threshold, it is determined that there is a long-range relationship between the two target entities, and corresponding search knowledge is obtained based on the two target entities and the long-range relationship.

[0108] In the embodiment of the present invention, the association relationship between entities is quantitatively judged through the minimum number of hops and the number of paths between entities, which can mine richer entity relationships and provide research data for the construction of subsequent full-factor mineral resource prediction models.

[0109] Step A2: Integrate the search knowledge to obtain the first mined knowledge.

[0110] In this embodiment, all search knowledge obtained through knowledge search technology is sorted out based on the attributes of the entity to obtain the corresponding first mining knowledge. Specifically, when obtaining the target entity given by the user, this embodiment uses knowledge retrieval technology to quickly find the same or most similar knowledge data from the relevant knowledge graph, further mine the strong and indirect entity relationships between entities, and identify the long-range and implicitly connected entity relationships between entities, which can provide data indication and reference content for the subsequent construction of a full-factor mineral resource prediction model.

[0111] Specifically, when the second knowledge data is the second mined knowledge, the above step S203 includes:

[0112] Step B1, constructing a data set based on the first knowledge data and / or the knowledge graph, the data set includes: a training data set and a test data set.

[0113] Step B2: Train the training data set through a preset knowledge graph embedding model to obtain a corresponding knowledge reasoning model.

[0114] In this embodiment, the specific content of the preset knowledge graph embedding model is not limited here, and it is adaptively adjusted based on actual needs. For example, the preset knowledge graph embedding model is a translation distance model, such as TransE, TransH, TransR, TransD and other series models; or the preset knowledge graph embedding model is a semantic matching model, such as the DistMult model, which is only used as an exemplary illustration.

[0115] Step B3: input the test data set into the knowledge reasoning model for reasoning to obtain reasoning knowledge.

[0116] Step B4: Integrate the inference knowledge to obtain the second mined knowledge.

[0117] The embodiment of the present invention performs knowledge reasoning based on existing knowledge data and utilizes a knowledge graph embedding model to mine new possible entity relationships, namely, reasoning knowledge, and provides research data for the subsequent construction of a full-factor mineral resource prediction model.

[0118] It should be noted that the traditional mineral resource prediction model based on knowledge graph often uses artificial qualitative probability through prediction factors, such as setting the probability of different factors to 0, 1 or 0.5 based on subjective evaluation; or the result driven by local regional data is used as the probability of the factor in the prediction model to participate in the calculation of weight. This type of method has a certain degree of subjectivity and cannot well reflect the actual impact of different factors on mineralization. Therefore, when the second knowledge data in this embodiment is the third mining knowledge, the above-mentioned step S203 designed includes:

[0119] Step C1, obtain all paths that associate the mineral resources with any tail entity in the first knowledge data and / or the knowledge graph.

[0120] Step C2, counting the occurrence frequency of different entity relationships in each path, calculating the relationship probability between mineral resources and tail entities, and determining the intensity index of the corresponding path based on the occurrence frequency and relationship probability.

[0121] Step C3: Integrate the strength indexes of all paths to obtain the third mined knowledge.

[0122] The embodiment of the present invention takes into account the different probabilities of different elements, minerals and rocks in affecting the mineralization of mineral resources, and determines the intensity index of the corresponding path based on the frequency of occurrence and relationship probability of the association path between entities. The intensity index is used to characterize the strength of the association relationship between the corresponding entities. It is possible to calculate new implicit knowledge based on graph analysis and perform knowledge mining with different weights on it, thereby providing research data for the subsequent construction of a full-factor knowledge graph-based mineralization prediction model.

[0123] Step S204: construct a mineral resource prediction model based on the first knowledge data and the second knowledge data. Figure 1 Step S104 of the illustrated embodiment will not be described in detail here.

[0124] In a specific embodiment, taking porphyry copper mine as an example, knowledge retrieval based on graph computing, data statistics and association analysis, knowledge analysis and knowledge reasoning based on machine learning are used to perform corresponding knowledge mining to achieve the discovery of various relationships such as indirect, fuzzy and composite (discovering weakly associated and indirectly associated entities, identifying long-range associations and implicit dependencies between entities), and the full-factor construction of porphyry copper mine prediction is achieved through knowledge-driven and data-driven aspects. Specifically, the detailed construction process of the intelligent construction system of the full-factor prediction index system based on the mineral resource knowledge graph includes:

[0125] Step D1: knowledge data mining of long-range associations and implicit associations based on knowledge graph search.

[0126] In this embodiment, a corresponding knowledge query method is developed based on the currently stored knowledge graph of porphyry copper deposits to realize various knowledge queries such as indirect, fuzzy, and composite. Specifically, the two query methods are as follows:

[0127] 1. Human-computer interactive query interface based on query method and parameter selection.

[0128] In this embodiment, six query modes for indirect, fuzzy, and composite relationships between entities are set, including: (1) N paths between entity A and entity B in LR hops; (2) N acyclic paths between entity A and entity B in LR hops; (3) the shortest path between entity A and entity B in LR hops; (4) the longest path between entity A and entity B in LR hops; (5) N paths between entity A and entity B in LR hops through entity C; (6) the maximum weight path between entity A and entity B in LR hops. Among them, parameter L represents the minimum number of hops required to associate two entities, parameter R represents the maximum number of hops required to associate two entities, and parameter N represents the number of paths that two entities have associated.

[0129] Specifically, the user queries the corresponding results by inputting the restrictions on the number of hops and paths required to associate two entities. Among them, the user can set the minimum or maximum number of hops for the association of two entities, as well as the number of associated paths. Specifically, the number of hops increases from small to large, and the number of paths increases from few to many, which can realize the gradual mining and analysis of relationship links from short to long, from direct to indirect, and from simple to complex. Based on the above rules, the mining results such as the discovery of strongly associated and indirectly associated entities, and the identification of long-range associations and implicit connections between entities provide instructions and references for the construction of porphyry copper deposit prediction elements.

[0130] 2. Provide users with an interface to query the knowledge graph by writing query language.

[0131] In this embodiment, the user uses the query language to write the corresponding query statement (such as SQL statement) according to the information and conditions to be queried; then the query statement is sent to the public knowledge graph interface to obtain the query results. Specifically, the query language can realize flexible query of the knowledge graph and obtain the required information, and realize the effective management and utilization of the knowledge graph data; at the same time, the standardized knowledge graph interface provided can enable different systems and applications to interact and integrate with the knowledge graph conveniently.

[0132] In a specific embodiment, Figure 3 This is a schematic diagram of the results of the knowledge graph query. Through the query interface, the entity relationship results of the two entities to be analyzed, namely, the user sets entity 1 copper element and entity 2 sulfur element, and the path between the two through entity 3 sulfide, where the hop count range (LR) is set to 0-3 and the number of paths (N) is set to 10. Specifically, Figure 3It can be seen that the behavior of copper and sulfur in magma is closely related to magnetite and hematite. This result can provide research direction and indication for exploring the formation mechanism of porphyry copper deposits. After further expert analysis and map mining, different oxygen fugacity conditions can make magnetite and hematite transform into each other, which provides a new long-range correlation discovery between oxygen fugacity and porphyry copper deposits, that is, magnetite can be used as an indicator mineral of porphyry copper deposits as a predictive factor.

[0133] Step D2: knowledge data mining based on knowledge reasoning and knowledge completion of the knowledge graph embedding model.

[0134] In this embodiment, knowledge completion or link prediction is implemented based on the Knowledge Graph Embedding (KGE) model, that is, whether a certain entity has a specific relationship with another given entity is inferred, so as to achieve further knowledge discovery. The specific process includes:

[0135] 1. Model selection.

[0136] In order to embed entities and relationships in the knowledge graph, this embodiment uses the bilinear ComplEx model for subsequent knowledge reasoning. Specifically, the model is very advantageous in solving 1-N, N-1, NN entity relationships, and can solve symmetric and asymmetric relationships. Its implementation capability is better than that of neural network models, bilinear models such as RESCAL and DisMult. In addition, in terms of the model's embedding expression capability, the model can expand the expression of entities and relationships from a simple tensor space domain to a complex complex space domain, and define a scoring function.

[0137] 2. Dataset preprocessing.

[0138] In this embodiment, the original data set (i.e., the first knowledge data and the knowledge graph) is divided into a training set, a validation set, and a test set, and then the corresponding data preprocessing is performed, and an index is assigned to each different entity name and each different relationship name. It should be noted that the specific method of data preprocessing is not limited here, and it is adaptively adjusted based on actual needs, such as integrating the data, including data information extraction and fusion, missing value and outlier processing, study area time slicing processing, spatial mapping and rasterization processing, etc., which are only exemplary.

[0139] 3. Initialize entity and relationship embeddings.

[0140] In this embodiment, the embedding of entities and entity relationships is randomly initialized according to the Gaussian distribution. In practical applications, if the number of samples in the data set is small, the number of samples can be enriched by generating negative training examples. Specifically, negative training examples are generated by randomly destroying positive training examples or by random sampling. Among them, in order to reduce the generation of false negative training examples, this embodiment also sets different probabilities of replacing head and tail entities, that is, if the relationship is 1-N, more opportunities are given to replace the head entity, and if the relationship is N-1, more opportunities are given to replace the tail entity. The generation quality of negative training examples is guaranteed by the above method.

[0141] 4. Model optimization and evaluation.

[0142] In this embodiment, the stochastic gradient descent algorithm is used to optimize the model. This algorithm has a great advantage in reducing computing resources. Ultimately, the optimal solution is obtained by adjusting the learning rate and continuously evaluating the model.

[0143] 5. Model application reasoning.

[0144] In this embodiment, the KGE model is used to infer the long-range relationships in the knowledge graph of porphyry copper deposits. First, the training results are loaded, the triples to be predicted are input, and the link scores of the tail entities are performed using the scoring function, wherein the higher the score, the more suitable it is. The link scoring results are then sorted by size, and the prediction results are interpreted by experts from top to bottom until the experts determine that the predicted relationship is almost uninterpretable. In this way, a threshold is set to ensure the rationality of the newly discovered long-range relationships.

[0145] In a specific embodiment, the relationship between the ridge subduction entity and the plate melting entity predicted by the knowledge graph embedding model KGE is highly probable (i.e., the predicted link relationship score is 0.93), indicating that there may be a direct causal relationship between them. This relationship does not exist in the existing knowledge graph and is obtained through the KGE model prediction reasoning calculation. Figure 4 is a schematic diagram of the result of knowledge reasoning. Figure 4 It can be seen that there is no direct or indirect relationship between the two circled entities, the subduction of the ocean ridge and the melting of the plate. The KGE model predicts that the probability is relatively high. There is some invisible connection between the two entities, so the two entities are connected through the arrow curve.

[0146] Step D3, knowledge data mining based on implicit knowledge calculated by graph analysis.

[0147] Based on the existing porphyry copper deposit knowledge graph, this embodiment realizes the prediction factor weight calculation applicable to the universality and interpretability of porphyry copper deposit prediction through dual drive of knowledge and data. The specific calculation process includes:

[0148] 1. Retrieve all the association paths between "porphyry copper deposits PCDs" and specific entities (such as "quartz") in the knowledge graph.

[0149] 2. According to the frequency of occurrence of the relationship between different entities in each path, the probability of the relationship between "porphyry copper deposits PCDs" and the specific entity is calculated. Specifically, this probability can be predicted by the aforementioned knowledge graph embedding model KGE.

[0150] 3. Calculate the joint probability of all paths as the strength index of the associated path to characterize the strength of the relationship between "porphyry copper PCDs" and specific entities. The strength of these relationships can be used as the calculation weight given to the element when participating in the prediction of mineral resource prediction models (such as fuzzy evidence weight model, artificial intelligence prospecting prediction model, etc.). Figure 5 It is a schematic diagram of calculation based on graph analysis.

[0151] Step D4, construction of a full-factor prediction model for porphyry copper deposits based on the knowledge graph.

[0152] Figure 6 This is a schematic diagram of the structure of the full-factor prediction model for porphyry copper deposits. Figure 6 It can be seen that in this embodiment, the full-factor prediction model of porphyry copper deposits is constructed through knowledge-driven and data-driven aspects. Among them, knowledge-driven is based on expert knowledge and theoretical models. By constructing a porphyry copper deposit knowledge system, it further summarizes and condenses the porphyry copper deposit mineralization prediction factor framework model containing four systems of "Earth system-metallogenic system-exploration system-prediction and evaluation system", such as plate tectonic elements, rock combination characteristic elements, indicator element combination elements, indicator mineral combination elements, etc. Specifically, the corresponding plate tectonic elements, such as subduction angle, subduction rate, and subduction sediment thickness, can be extracted based on Gplate (which is a software for plate tectonic simulation and earth evolution research); rock combination characteristic elements such as oxygen fugacity and water content of magma are obtained based on rock and mineral chemical data; indicator mineral combination elements such as rock type and ore-controlling structure are obtained based on geological mapping; indicator mineral combination elements such as indicator minerals and alteration zoning extraction based on remote sensing; indicator mineral combination elements such as chemical element anomaly combination based on geochemical mapping; indicator mineral combination elements such as deep concealed rock body identification based on geophysics.

[0153] In this embodiment, data-driven is based on the constructed porphyry copper knowledge graph and prediction full-factor framework. Through the aforementioned spectrum search, spectrum analysis, knowledge reasoning and other knowledge discovery technologies on the porphyry copper knowledge graph, the direct, indirect, long-range and implicit element indicators within the element framework are mined, and the relationship strength index of each element indicator is calculated, aiming to maximize the reflection of the output mechanism and prospecting signs of porphyry copper, and then the scientific data and engineering exploration data corresponding to the full-factor model are mined and processed to obtain the corresponding new knowledge data. Finally, through the integration and processing of all-factor data, including data information extraction and fusion, missing value and outlier processing, research area time slicing processing, spatial mapping and rasterization processing, etc., the artificial intelligence model is used to complete the accurate artificial intelligence prediction of porphyry copper deposits in typical mineralization domains.

[0154] The embodiment of the present invention regards porphyry copper deposits as the product of the earth's evolution process. By considering the relationship between it and the multi-sphere system inside the earth, the "earth system" is included in the mineral prediction research, and the complex connection between it and the mineralization system, exploration system and prediction and evaluation system is constructed, which will strengthen its research and understanding of the remote dynamic association and mutual restriction of various mineralization-related factors; at the same time, the cutting-edge technology in machine learning and deep learning is used to construct a data mining algorithm to mine hidden factors related to porphyry copper deposits from high-dimensional data, and provide new technical solutions and references for knowledge graph analysis and mining in various fields. The construction of the full-factor prediction model of porphyry copper deposits based on the knowledge graph in this embodiment can reflect the output mechanism and prospecting signs of porphyry copper deposits to the greatest extent, and use this as a guide for the processing of engineering detection data, so as to achieve the purpose of accurate prediction.

[0155] In summary, the knowledge graph-based mineralization prediction model construction method of the embodiment of the present invention can take into account the influence of factors such as the lithospheric material structure, crust-mantle evolution, and the evolution and circulation of mineralizing materials on mineralization, realize the comprehensive utilization of mineral resource elements, improve the accuracy of prediction results, and meet the intelligent requirements of mineral resource prediction.

[0156] In this embodiment, a system for building a mineralization prediction model based on a knowledge graph is also provided, which is used to implement the above-mentioned embodiments and preferred embodiments, and the descriptions that have been made will not be repeated. As used below, the term "module" refers to a combination of software and / or hardware that can implement a predetermined function. Although the system described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceived.

[0157] The present invention provides a system for constructing a mineralization prediction model based on a knowledge graph, such as Figure 7 As shown, the system includes:

[0158] The data construction module 701 is used to construct a knowledge graph of mineral resources, which contains knowledge data of elements, minerals, rocks, ore deposits, mineralization and geological phenomena.

[0159] The knowledge extraction module 702 is used to extract knowledge from the knowledge graph based on the data types of the earth system, mineralization system, exploration system and prediction and evaluation system to obtain the first knowledge data.

[0160] The knowledge mining module 703 is used to perform knowledge mining on the first knowledge data and / or knowledge graph based on a preset knowledge mining method to obtain second knowledge data.

[0161] The model building module 704 is used to build a mineral resource prediction model based on the first knowledge data and the second knowledge data.

[0162] In some optional embodiments, the data construction module 701 includes: a first acquisition submodule, a second acquisition submodule, a third acquisition submodule, a fourth acquisition submodule and a fifth acquisition submodule; wherein the first acquisition submodule is used to collect relevant data on mineral resources including earth systems, mineralization systems, exploration systems and prediction and evaluation systems; the second acquisition submodule is used to annotate entities and relationships of relevant data to obtain a large amount of knowledge data, wherein the knowledge data is triple data including entities and entity relationships, wherein entities represent elements related to mineralization of mineral resources, and entity relationships represent semantic relationships between different entities; the third acquisition submodule is used to extract training data and test data from a large amount of knowledge data, and to construct a data generation model based on the training data, wherein the data generation model is used to generate triple data including entities and entity relationships; the fourth acquisition submodule is used to input test data into the data generation model for prediction, and correspondingly obtain a large amount of generated triple data; the fifth acquisition submodule is used to store all triple data to obtain a knowledge graph of mineral resources.

[0163] In some optional embodiments, the knowledge mining module 703 includes: a first search submodule and a second search submodule; wherein the first search submodule is used to respond to two target entities selected by the user from the first knowledge data and / or knowledge graph, and to perform knowledge search on the two target entities based on a preset entity relationship type query method to obtain corresponding search knowledge; the second search submodule is used to integrate the search knowledge to obtain the first mined knowledge.

[0164] In some optional embodiments, the first search submodule includes: a first search unit, a second search unit, a third search unit, a fourth search unit, a fifth search unit and a sixth search unit; wherein the first search unit is used to obtain the minimum number of hops and the number of paths required for the association between the two target entities; the second search unit is used to determine whether the minimum number of hops is less than a preset hop threshold; the third search unit is used to determine that the two target entities have a direct relationship if the minimum number of hops is less than the preset hop threshold, and obtain corresponding search knowledge based on the two target entities and the direct relationship; the fourth search unit is used to determine that the two target entities have an indirect relationship if the minimum number of hops is not less than the preset hop threshold, and determine whether the number of paths is less than the preset path threshold; the fifth search unit is used to determine that the two target entities have a short-range relationship if the number of paths is less than the preset path threshold, and obtain corresponding search knowledge based on the two target entities and the short-range relationship; the sixth search unit is used to determine that the two target entities have a long-range relationship if the number of paths is not less than the preset path threshold, and obtain corresponding search knowledge based on the two target entities and the long-range relationship.

[0165] In some optional embodiments, the knowledge mining module 703 includes: a first reasoning sub-module, a second reasoning sub-module, a third reasoning sub-module and a fourth reasoning sub-module; wherein the first reasoning sub-module is used to construct a data set based on the first knowledge data and / or knowledge graph, and the data set includes: a training data set and a test data set; the second reasoning sub-module is used to train the training data set through a preset knowledge graph embedding model to obtain a corresponding knowledge reasoning model; the third reasoning sub-module is used to input the test data set into the knowledge reasoning model for reasoning to obtain reasoning knowledge; the fourth reasoning sub-module is used to integrate the reasoning knowledge to obtain the second mined knowledge.

[0166] In some optional embodiments, the knowledge mining module 703 includes: a first calculation submodule, a second calculation submodule and a third calculation submodule; wherein the first calculation submodule is used to obtain all paths in which mineral resources are associated with any tail entity in the first knowledge data and / or the knowledge graph; the second calculation submodule is used to count the occurrence frequency of different entity relationships in each path, calculate the relationship probability between mineral resources and the tail entity, and determine the strength index of the corresponding path based on the occurrence frequency and the relationship probability; the third calculation submodule is used to integrate the strength index of all paths to obtain the third mined knowledge.

[0167] The further functional description of each of the above modules is the same as that of the above corresponding embodiments and will not be repeated here.

[0168] The knowledge graph-based mineralization prediction model construction system in this embodiment is presented in the form of functional units, where the units refer to ASIC (Application Specific Integrated Circuit) circuits, processors and memories that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0169] The knowledge graph-based mineralization prediction model construction system of the embodiment of the present invention takes into account the influence of factors such as the lithospheric material structure, crust-mantle evolution, and the evolution and circulation of ore-forming materials on mineralization, and obtains corresponding knowledge data based on the earth system data type, mineralization system data type, exploration system data type, and prediction and evaluation system data type, and further mines new knowledge data from the data to construct a mineral resource prediction model, thereby realizing the comprehensive utilization of mineral resource elements, helping to improve the accuracy of prediction results, and meeting the intelligent requirements of mineral resource prediction.

[0170] The embodiment of the present invention also provides a computer device, see Figure 8 , Figure 8 is a schematic diagram of the structure of the above-mentioned computer device provided by an optional embodiment of the present invention, such as Figure 8 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 8 A processor 10 is taken as an example.

[0171] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0172] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0173] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0174] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0175] The computer device also includes a communication interface 30, which is used for the main control chip to communicate with other devices or a communication network.

[0176] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor main control chip or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor or hardware, the method shown in the above embodiment is implemented.

[0177] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0178] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for constructing a mineralization prediction model based on a knowledge graph, characterized in that: The method comprises: Constructing a knowledge graph of mineral resources, the knowledge graph comprising knowledge data of elements, minerals, rocks, ore deposits, mineralization and geological phenomena; Extracting knowledge from the knowledge graph based on the data types of the earth system, the mineralization system, the exploration system, and the prediction and evaluation system to obtain first knowledge data; Performing knowledge mining on the first knowledge data and / or the knowledge graph based on a preset knowledge mining method to obtain second knowledge data; A mineral resource prediction model is constructed based on the first knowledge data and the second knowledge data.

2. The method for constructing a mineralization prediction model based on a knowledge graph according to claim 1, characterized in that: The construction of the knowledge graph of mineral resources includes: Collect relevant data on mineral resources including earth system, mineralization system, exploration system and prediction and evaluation system; Perform entity and relationship annotation on the relevant data to obtain a large amount of knowledge data, wherein the knowledge data is triple data including entities and entity relationships, wherein the entities represent elements related to the mineralization of the mineral resources, and the entity relationships represent semantic relationships between different entities; Extracting training data and test data from a large amount of knowledge data, and constructing a data generation model based on the training data, wherein the data generation model is used to generate triple data including entities and entity relationships; The test data is input into the data generation model for prediction, and a large amount of generated triple data is obtained accordingly; All triplet data are stored to obtain a knowledge graph of the mineral resources.

3. The method for constructing a mineralization prediction model based on a knowledge graph according to any one of claims 1 to 2, characterized in that: The second knowledge data includes first mined knowledge, and the first knowledge data and / or the knowledge graph are mined based on a preset knowledge mining method to obtain the second knowledge data, including: In response to two target entities selected by the user from the first knowledge data and / or the knowledge graph, knowledge search is performed on the two target entities based on a preset entity relationship type query method to obtain corresponding search knowledge; The search knowledge is integrated to obtain first mined knowledge.

4. The method for constructing a mineralization prediction model based on a knowledge graph according to claim 3, characterized in that: The preset entity relationship type includes a direct relationship and an indirect relationship, wherein the indirect relationship includes a short-range relationship and a long-range relationship; the knowledge search is performed on the two target entities based on the preset entity relationship type query method to obtain the search knowledge, including: Get the minimum number of hops and paths required for the two target entities to be associated; Determine whether the minimum hop count is less than a preset hop count threshold; If the minimum hop count is less than a preset hop count threshold, determining that the two target entities have a direct relationship, and obtaining corresponding search knowledge based on the two target entities and the direct relationship; If the minimum hop count is not less than a preset hop count threshold, determining that the two target entities have an indirect relationship, and determining whether the number of paths is less than a preset path threshold; If the number of paths is less than a preset path threshold, determining that there is a short-range relationship between the two target entities, and obtaining corresponding search knowledge based on the two target entities and the short-range relationship; If the number of paths is not less than a preset path threshold, it is determined that there is a long-range relationship between the two target entities, and corresponding search knowledge is obtained based on the two target entities and the long-range relationship.

5. The method for constructing a mineralization prediction model based on a knowledge graph according to any one of claims 1 to 2, characterized in that: The second knowledge data includes second mined knowledge, and the first knowledge data and / or the knowledge graph are mined based on a preset knowledge mining method to obtain the second knowledge data, including: Constructing a data set based on the first knowledge data and / or the knowledge graph, the data set comprising: a training data set and a test data set; The training data set is trained through a preset knowledge graph embedding model to obtain a corresponding knowledge reasoning model; Inputting the test data set into the knowledge reasoning model for reasoning to obtain reasoning knowledge; The inference knowledge is integrated to obtain second mined knowledge.

6. The method for constructing a mineralization prediction model based on a knowledge graph according to any one of claims 1 to 2, characterized in that: The second knowledge data includes third mined knowledge, and the first knowledge data and / or the knowledge graph are mined based on a preset knowledge mining method to obtain the second knowledge data, including: Obtain all paths where mineral resources are associated with any tail entity in the first knowledge data and / or the knowledge graph; Counting the occurrence frequencies of different entity relationships in each path, calculating the relationship probability between the mineral resources and the tail entity, and determining the intensity index of the corresponding path based on the occurrence frequencies and the relationship probability; The strength indexes of all paths are integrated to obtain the third mining knowledge.

7. A system for building a mineralization prediction model based on knowledge graph, characterized in that: The system comprises: A data construction module, used to construct a knowledge graph of mineral resources, wherein the knowledge graph contains knowledge data of elements, minerals, rocks, ore deposits, mineralization and geological phenomena; A knowledge extraction module, used to extract knowledge from the knowledge graph based on the data types of the earth system, the mineralization system, the exploration system and the prediction and evaluation system to obtain first knowledge data; A knowledge mining module, configured to perform knowledge mining on the first knowledge data and / or the knowledge graph based on a preset knowledge mining method to obtain second knowledge data; A model building module is used to build a mineral resource prediction model based on the first knowledge data and the second knowledge data.

8. A computer device, characterized in that: The device includes: a memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method for constructing a mineralization prediction model based on a knowledge graph as described in any one of claims 1 to 6 by executing the computer instructions.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method for constructing a mineralization prediction model based on a knowledge graph as described in any one of claims 1 to 6.

10. A computer program product, characterized in that It includes computer instructions, which are used to enable a computer to execute the method for constructing a mineralization prediction model based on a knowledge graph as described in any one of claims 1 to 6.

Citation Information

Patent Citations

  • Geological knowledge graph construction method and device, electronic equipment and storage medium

    CN114218333A

  • Phosphorite metallogenic rule text data mining method and system

    CN116089629A

  • Mapping knowledge domain-based overseas copper mine potential prediction method and system, and electronic equipment

    CN116090662A

  • Mineral resource prediction method based on knowledge graph driving and storage medium

    CN116307123A

  • Entity type reasoning method, system and equipment and storage medium

    CN116821364A

Cited By

  • Sintering dynamic knowledge graph construction method

    CN119494395A

  • Big data intelligent prospecting operation method based on multi-source geological relation data

    CN119740808A

  • Mineral resource quantitative prediction method and system based on knowledge embedding

    CN119783870A