Method and device for extracting geological prospecting knowledge triples using prompt word engineering

Through a large language model combining the ontology model of the mineral exploration knowledge graph and prompt word engineering, geological mineral exploration knowledge triplets are extracted in step by step, solving the problem of time-consuming and labor-consuming manual labeling in traditional methods, and achieving efficient and accurate triplet data extraction and management.

CN120087465BActive Publication Date: 2025-08-22CHINA GEOLOGICAL LIBRARY (GEOLOGICAL LITERATURE CENT OF CHINA GEOLOGICAL SURVEY)
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202510211498.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-02-25
Publication Date
2025-08-22
Estimated Expiration
2045-02-25

AI Technical Summary

Technical Problem

Traditional machine learning to extract geological prospecting knowledge triplets requires a lot of manual annotation, and the existing methods are inefficient, resulting in insufficient comprehensiveness and accuracy of the output triplets.

Method used

The large language model is used to combine the ontology model of the mineral exploration knowledge graph, and the geological mineral exploration knowledge triplets are extracted in step by step through the prompt word engineering, including defining entity types and relationship types, analyzing and formatting the triplets paragraph by paragraph, and optimizing the prompt word to improve the extraction accuracy.

Benefits of technology

It reduces the cost of manual labeling, improves the efficiency and accuracy of triple data extraction, systematically manages geological prospecting knowledge, and provides strong technical support.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120087465B_ABST
    Figure CN120087465B_ABST
Patent Text Reader

Abstract

The present invention provides a method and apparatus for extracting geological prospecting knowledge triples using prompt word engineering. The method comprises: defining a prospecting knowledge graph ontology model; the prospecting knowledge graph ontology model includes multiple entity types and multiple relationship types; and, in conjunction with the prospecting knowledge graph model, using prompt word engineering to step-by-step extract geological prospecting knowledge triples from a target file. By integrating the prospecting knowledge graph ontology model, the present invention can learn multiple features of geological text information, improving the accuracy of knowledge extraction. By employing extraction engineering, entities and attributes related to geological prospecting can be automatically extracted, significantly reducing labor costs and improving the efficiency of triple data extraction.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of computer natural language processing, in particular to a method and device for extracting geological prospecting knowledge triples by using prompt word engineering. Background Art

[0002] A triple is a set of information consisting of an entity, a relation, and an attribute that together describe a specific fact in geological data. Traditional machine learning methods for extracting triples require extensive manual annotation, and using them as machine learning corpus is labor-intensive.

[0003] In addition, when existing technologies extract triplet data through machine learning, they usually input prompt words once based on the entire document, which makes the entire process long and inefficient, and the output triplet is slightly lacking in comprehensiveness and accuracy. Summary of the Invention

[0004] In order to solve the above problems, the purpose of the present invention is to provide a method and device for extracting geological prospecting knowledge triples using prompt word engineering. The present invention uses a large language model to automatically extract knowledge triples, without the need for a large amount of manual labeling in the early stage, saving labor costs; by optimizing the existing prompt words, the prompt words are refined and divided into steps, and the prompt words are executed step by step, thereby improving the extraction accuracy.

[0005] The present invention provides a method for extracting geological prospecting knowledge triples using prompt word engineering, the method comprising:

[0006] Defining a prospecting knowledge graph ontology model; the prospecting knowledge graph ontology model includes multiple entity types and multiple relationship types;

[0007] Combined with the prospecting knowledge graph model, the prompt word engineering is used to extract the geological prospecting knowledge triples of the target file step by step.

[0008] Optionally, the entity types include a first entity type for basic geological information, a second entity type for gold prospecting, and a third entity type for studying the genesis of mineral deposits;

[0009] The first entity type includes regional geological background or mineral deposit geological background, including region, mineral belt, mineral cluster, mineral deposit, mineral section, vein, stratum, structure, rock mass, mineralization type, alteration, formation age, geological event, mineralization mechanism, mineralization stage, reserves and other characteristics; the other characteristics include geophysical characteristics, geochemical characteristics, ore-controlling structural characteristics, alteration zoning, mineralization zoning, remote sensing image characteristics, isotopic composition, mineral composition, ore structure / structure, mining area characteristics, vein characteristics, rock composition, minerals, grade, homogenization temperature, strike, dip, length, width and thickness;

[0010] The second entity type includes prospecting target areas, remote sensing interpretation anomalies, geophysical anomalies, and geochemical anomalies;

[0011] The third entity type includes: inclusions, material sources, and mineralization conditions.

[0012] Optionally, in combination with the prospecting knowledge graph model, the geological prospecting knowledge triples of the target file are extracted step by step using the prompt word engineering, including:

[0013] Parsing the target file into text in article format and segmenting the text into paragraphs;

[0014] Combined with the prospecting knowledge graph model, the prompt word engineering is used to extract the geological prospecting knowledge triples of each paragraph step by step;

[0015] Merge the triple extraction results corresponding to each paragraph into a table.

[0016] Optionally, in combination with the prospecting knowledge graph model, the prompt word engineering is used to extract the geological prospecting knowledge triples of each paragraph step by step, including:

[0017] Extract the triples in the current paragraph;

[0018] Processing the triples to resolve unclear entity references in the triples;

[0019] Determine and filter the head entity type and tail entity type;

[0020] Summarizing the head entity type and the tail entity type;

[0021] Determine and filter the relationship between the head entity and the tail entity;

[0022] Format a list of triplets into a table;

[0023] Deduplication of the table;

[0024] Save this output as the current paragraph extraction result.

[0025] Prompt Engineering is a discipline that develops and optimizes prompts to help users effectively apply language models to various application scenarios and research areas. In triple extraction models, prompt engineering can be used to improve model performance and efficiency.

[0026] Optionally, extracting triples from the current paragraph content includes:

[0027] Sentence-by-sentence analysis: The document paragraph is broken down sentence by sentence, and independent triples are extracted for each sentence. This includes: identifying the head entity: finding the main object or concept discussed in the sentence; determining the relationship: analyzing the relationship between the head entity and other components in the sentence; and locating the tail entity: finding another object or concept connected to the head entity through a relationship.

[0028] Record the original text: Attach the original sentence from the original text to each extracted triple for subsequent verification and screening;

[0029] Consideration of inter-sentence relations: After completing sentence-by-sentence extraction, review the logical relations between sentences and check whether there are triplets across sentences that can be extracted.

[0030] Optionally, processing the triple to solve the problem of unclear entity reference in the triple includes:

[0031] Identify ambiguous entities: Review the initially extracted triples and identify entities with ambiguous references;

[0032] Contextual analysis: For each unidentified entity, go back to the original paragraph and analyze the context in which it appears to determine its exact reference;

[0033] Entity association: Based on contextual information, the entity with unknown reference is associated with the entity clearly mentioned in the text;

[0034] Update triples: Update triples based on the clarified entity references to ensure that each entity is clear and unambiguous.

[0035] Optionally, determining and filtering the head entity type and the tail entity type includes:

[0036] Entity non-empty verification: Check whether the head entity and tail entity in each triple are empty. If either entity is empty, delete the corresponding triple;

[0037] Entity type determination: After confirming that the entity is not empty, the entity type is determined; the head entity and the tail entity are classified according to the preset entity type list. If the entity type does not belong to any type in the entity type list, the corresponding triple is deleted.

[0038] Optionally, determining and filtering the relationship between the head entity and the tail entity includes:

[0039] Relationship screening: Check whether the relationship in each triple can be mapped to one of the preset relationship lists;

[0040] Relationship determination: Based on the original content, determine whether the relationship in the triple can be accurately converted to any relationship in the relationship list; if the relationship is not in the preset relationship list, or cannot be accurately mapped to any relationship in the relationship list based on the original content, mark it as invalid and delete the triple.

[0041] Optionally, formatting the list of triplets into a table involves:

[0042] Create a table: Use the following column headings to construct the table: - Sequence number - Head entity summary type - Head entity type - Head entity - Relationship between entities - Tail entity summary type - Tail entity type - Tail entity - Relationship attributes - Other attributes - Original text - Article number - Remarks;

[0043] Fill in the form information.

[0044] The present invention provides a device for extracting geological prospecting knowledge triples using prompt word engineering, comprising a processor and a memory storing program instructions, characterized in that the processor is configured to execute the method for extracting geological prospecting knowledge triples using prompt word engineering as described in any of the above items when executing the program instructions.

[0045] The method and device provided by this invention for extracting triples of geological prospecting knowledge using prompt word engineering, combined with a prospecting knowledge graph ontology model, can learn the various characteristics of geological text information, improving the accuracy of knowledge extraction. By using extraction engineering, entities and attributes related to geological prospecting can be automatically extracted, enabling systematic knowledge management, intelligent extraction, and knowledge integration. This significantly reduces labor costs and improves the efficiency of triple data extraction, providing strong technical support for geological prospecting.

[0046] Based on the following detailed description of specific embodiments of the present invention in conjunction with the accompanying drawings, those skilled in the art will become more aware of the above and other objects, advantages and features of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0048] Figure 11 is a flow chart of a method for extracting geological prospecting knowledge triples using prompt word engineering according to an embodiment of the present invention;

[0049] Figure 2 2. It is a schematic diagram of the prospecting knowledge graph ontology model according to an embodiment of the present invention;

[0050] Figure 3 This is a schematic diagram of basic knowledge of gold mines according to an embodiment of the present invention;

[0051] Figure 4 2 is a schematic diagram of step-by-step triple extraction according to an embodiment of the present invention. DETAILED DESCRIPTION

[0052] The embodiments of the present invention are described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to illustrate the present invention and are not intended to limit the present invention.

[0053] The present invention provides a method for extracting geological prospecting knowledge triples using prompt word engineering, such as Figure 1 As shown, the method includes:

[0054] S1, defining a prospecting knowledge graph ontology model; the prospecting knowledge graph ontology model includes multiple entity types and multiple relationship types;

[0055] S2, combining the prospecting knowledge graph model, using the prompt word engineering to extract the geological prospecting knowledge triples of the target file step by step.

[0056] The embodiment of the present invention first defines the prospecting knowledge graph ontology model. Figure 2 As shown, the embodiment of the present invention defines 89 entity types and 69 relationship types.

[0057] 1. Entity Type

[0058] Entity types are divided into the following three aspects: the first entity type for basic geological information, the second entity type for gold prospecting, and the third entity type for the research of the genesis of the ore deposits.

[0059] In the embodiment of the present invention, it mainly refers to the gold mine related knowledge. For gold mines, the basic knowledge of ore deposits mainly refers to the mineralization geological background or regional geological background, which may mainly include regional geology and mining area geology related content, such as Figure 3 Entity description and annotation specifications: By analyzing multi-source heterogeneous mineral deposit resources, it is concluded that the relevant description information of the mineral deposit can be summarized as follows:

[0060] First, basic knowledge of the basic geological information of the deposit is introduced. This part includes the geological age of the formation of the gold mine, rock types, mineral types, and geological structures and geological activities related to prospecting.

[0061] Second, through previous research or reports, the mineral deposits and rocks in the mining area are classified, the content of relevant chemical components is calculated, the minerals are divided and classified, and the location of the mining area is delineated.

[0062] Third, based on the geological structure background, mineral position and reserves that have been explored by previous researchers, this section may include specific geophysical or geochemical method measurements and result inferences, which are also annotation content.

[0063] 1. The first entity type includes basic geological information, namely the regional geological background or the geological background of the mineral deposit; including regions, mineral belts, mineral clusters, mineral deposits, mineral sections, veins, strata, structures, rock bodies, mineralization types, alteration, formation age, geological events, mineralization mechanisms, mineralization stages, reserves, and other characteristics.

[0064] 1.1 Regions include: plateaus, basins, cratons, plates, rift valleys, subduction zones, suture zones, orogenic belts, collision zones, geographical locations, shields, terranes, platforms, fault zones, and alteration zones.

[0065] 1.1.1 Plateau

[0066] A plateau refers to a high plateau with a high altitude (generally greater than 500 meters), a vast area, a relatively low topography, and steeper outer surface slopes. A plateau is distinguished from a mountain by its vast, gently undulating terrain and relatively low surface relief, and from a plain by its greater altitude. The topography of plateaus is highly complex and can generally be divided into dissected plateaus and undulating plateaus. The four largest plateaus in the world are the Antarctic Glacial Plateau (12.8 million square kilometers), the Qinghai-Tibet Plateau (approximately 2.5 million square kilometers), the Iranian Plateau (2.5 million square kilometers), and the Kalahari Plateau (2.1 million square kilometers). China's four major plateaus are the Qinghai-Tibet Plateau, the Inner Mongolia Plateau, the Loess Plateau, and the Yunnan-Guizhou Plateau. The Qinghai-Tibet Plateau is a tall, mountainous plateau with a well-defined shape; the Inner Mongolia Plateau is a gently undulating plateau; the Yunnan-Guizhou Plateau is characterized by numerous high mountains and canyons; and the Loess Plateau has a fragmented surface due to intense erosion. The formation of plateau surfaces is complex and can be the result of long-term erosion and excavation, or accumulations of basalt flows. The main reason plateaus rise above the surrounding lowlands is due to long-term, continuous, and large-scale uplift. The rate of uplift varies from plateau to plateau. For example, since the Quaternary Period, the Qinghai-Tibet Plateau has risen by more than 3,000 meters, the Inner Mongolia Plateau by nearly 1,000 meters, and the Loess Plateau by a smaller margin of about 300 meters.

[0067] [Annotation example]

[0068] Original text: The West Qinling Orogenic Belt is located on the northeastern edge of the Qinghai-Tibet Plateau. It is the western extension of the Central Qinling Orogenic Belt of the Qin-Qi-Kun Central Orogenic Belt. It is sandwiched between the Qilian Orogenic Belt and the Songpan-Ganzi Orogenic Belt in the north and south. It is roughly bounded by the Baoji-Chengdu Railway in the east and the Qinghai Duohemao Fault in the west, which separates it from the Kunlun Orogenic Belt.

[0069] Triplet:

[0070] Head entity type: Orogenic belt

[0071] Head entity: West Qinling Orogenic Belt

[0072] Relationship: Located

[0073] Tail Entity Type: Plateau

[0074] Tail entity: Qinghai-Tibet Plateau

[0075]

[0076] 1.1.2 Basin

[0077] A basin-shaped depression surrounded by mountains, often with a central depression. Basins vary in size. The mountains surrounding large basins are often formed by folding and faulting, leaving the interior of the basin as a relatively stable or subsiding massif with a relatively flat surface. By origin: ① Tectonic basins, whose distribution and size are controlled by tectonic forces. ② Erosional basins, primarily controlled by external geological forces, include river valley basins, glacial basins, wind erosion basins, and karst basins. Basins can also be categorized by their geographical location as inland basins or external drainage basins. The central elevations of basins vary greatly. For example, the Qaidam Basin in western China reaches an elevation of 2,700 meters, while the Turpan Basin in Xinjiang, China, is at -155 meters, the lowest point on land in China. Many inland basins, surrounded by mountains, are difficult for moist maritime air to enter, resulting in scarce precipitation and an extremely dry climate. Examples include the Tarim Basin in Xinjiang, China, and the Great Basin in the western United States.

[0078] [Annotation example]

[0079] Original text: Jiaodong is located in eastern China, with its western part belonging to the North China Block, consisting of the Jiaobei Uplift and the Jiaolai Basin.

[0080] Triplet:

[0081] Header Entity Type: Geographic Location

[0082] Head entity: Jiaodong

[0083] Relationship: Composition

[0084] Tail entity type: Basin

[0085] Tail entity: Jiaolai Basin

[0086]

[0087] 1.1.3 Craton

[0088] A tectonic unit that has remained stable since the formation of the Earth's crust (at least since the Phanerozoic Eon) and has rarely undergone significant tectonic deformation. The term originally encompassed both continents and oceans. However, subsequent geological and geophysical data have confirmed the impossibility of craton structures within ocean basins, leading the term "craton" to now refer to continental cratons. Stille (1936) defined this type of tectonic unit as a region of tectonically stable areas since the Precambrian.

[0089] [Annotation example]

[0090] Original text: Jiaodong Peninsula is located on the eastern edge of the North China Craton

[0091] Triplet:

[0092] Header Entity Type: Geographic Location

[0093] Head entity: Jiaodong Peninsula

[0094] Relationship: Located

[0095] Tail entity type: Craton

[0096] Tail entity: North China Craton

[0097]

[0098] 1.1.4 Section

[0099] The Earth's lithosphere is divided into discontinuous lithosphere blocks of varying sizes by three major tectonic activity zones: mid-ocean ridges, island arc trench systems, and transform faults.

[0100] [Annotation example]

[0101] Original text: It is a post-collision granite formed by the subduction, collision, exhumation and extension of the Yangtze Plate and the North China Plate between 235 and 200 Ma.

[0102] Triplet:

[0103] Header entity type: Plate

[0104] Head entity: Yangtze Plate and North China Plate

[0105] Relationship: Formation

[0106] Tail entity type: igneous rock

[0107] Tail Entity: Granite

[0108]

[0109] 1.1.5 Rift

[0110] A narrow, long, linear depression bounded by high-angle normal faults. Rifts are the product of extensional tectonics, which thin and fracture the lithosphere. Once the crust is completely separated, new oceanic crust forms in between. Thus, they reflect the process of continental breakup and ocean basin formation. Rifts occur during the initial stages of the Wilson Cycle. There are two types of rifts: active rift valleys, which arise from the long-term action of hot mantle upwelling on the lithosphere's base, causing uplift, thinning, and fracture, such as the East African Rift. Passive rift valleys, in which crustal extension or shearing itself causes lithosphere thinning and fracture, triggering the passive upwelling of asthenospheric material, such as the Rhine Rift. In their evolutionary history, uplift and volcanic activity in active rifts precede subduction. The resulting rifts are often connected to global ridge systems. Passive rifts, on the other hand, follow the opposite evolutionary order. Rifts are characterized by their linear morphology, particularly the presence of alkaline bimodal volcanic complexes. In the ancient rift system of geological period, underlying cyclic alkaline complexes are also exposed, such as the Oslo Complex in Sweden. They are important signs for identifying rift systems.

[0111] [Annotation example]

[0112] Original text: In the Jiaodong area, the Jiaolai Basin, Tanlu Rift Valley and related normal faults were formed.

[0113] Triplet:

[0114] Head Entity Type: Rift

[0115] Head Entity: Tanlu Rift Valley

[0116] Relationship: Related

[0117] Tail Entity Type: Fault

[0118] Tail entity: normal fault

[0119]

[0120] 1.1.6 Subduction Zone

[0121] The subducting portion of a subducting plate. Plate margins where subduction occurs include B-type subduction zones (ocean-to-continent or ocean-to-ocean subduction) and A-type subduction zones (continent-to-continent subduction). B-type subduction zones feature large Benioff zones and are characterized by the development of trenches and arcs, as well as intense earthquakes and volcanic activity. They are generally composed of the following components: trenches formed by the downward bending of the subducting slab; forearc accretionary wedges scraped off by the subducting plate; enriched mantle wedges that overlie the plate's leading edge; volcanic arcs formed by partial melting of the slab at certain depths; and paired metamorphic belts associated with these arcs. Because oceanic lithosphere enters the mantle at subduction zones and is assimilated into mantle melt at certain depths, they are also called consuming boundary zones. B-type subduction zones are further divided into Chilean-type subduction zones (high-stress, gently dipping subduction zones with backarc compression) and Mariana-type subduction zones (low-stress, steeply dipping subduction zones with backarc expansion). A-type subduction zones are the product of the mutual subduction of continental lithosphere.

[0122] [Annotation example]

[0123] Original text: Magnesium diorite is generally believed to be produced in the arc basin system above the oceanic subduction zone.

[0124] Triplet:

[0125] Head entity type: igneous rock

[0126] Head entity: magnesian diorite

[0127] Relationship: Produced in

[0128] Tail entity type: subduction zone

[0129] Tail Entity: Oceanic Subduction Zone

[0130]

[0131] 1.1.7 Suture tape

[0132] Suture zones are areas where continental blocks are joined together by collisions of the Earth's crust. Suture zones contain various examples of modified sedimentary sequences associated with oceanic basins, rift continental margin sedimentary prisms, and arc-trench systems. They are typically mixed accumulations of ophiolite fragments and oceanic lithofacies. Furthermore, during crustal collisions, the margins of continental blocks sink due to partial subduction, forming marginal basins around the suture zone on the partially subducted plate. The most characteristic sediments in marginal basins are fluvial and deltaic strata, whose detritus originates from the continental margin. However, when the basin is deep, these deposits may be preceded by turbidite and deepwater sediments. (Yu Bingsong and Zhao Zhidan, Petrology, 2nd Edition, Beijing: Geological Publishing House, June 2012)

[0133] [Annotation example]

[0134] Original text: The Qinling orogenic belt is bounded by the Shangdan suture zone and the westward extension of the Tianshui-Wushan-Linxia fault.

[0135] Triplet:

[0136] Head entity type: Orogenic belt

[0137] Head entity: Qinling orogenic belt

[0138] Relationship: Boundary

[0139] Tail Entity Type: Suture Band

[0140] Tail entity: Shangdan suture

[0141]

[0142] 1.1.8 Orogenic belt

[0143] Tectonic belts on Earth with a zonal distribution and a distinct orogenic polarity are found at plate margins, between plates, or within plates. These belts are characterized by intense tectonic deformation and concentrated hydrothermal activity, making them crucial locations for studying the composition, structure, deformation, and geodynamics of the Earth's crust or lithosphere. Major types include the Andean type (oceanic crust on one side, continental crust on the other), the Himalayan type (continent-continent collision), and the New Guinea type (continent-island arc collision).

[0144] [Annotation example]

[0145] Original text: The Sulu orogenic belt is a high-pressure to ultrahigh-pressure metamorphic rock belt containing eclogite, representing the suture zone between the North China Block and the South China Block, and is part of the Jiaonan Block.

[0146] Triplet:

[0147] Head entity type: Orogenic belt

[0148] Head entity: Sulu orogenic belt

[0149] Relationship: Representative

[0150] Tail Entity Type: Suture Band

[0151] Tail entity: suture zone between North China and South China blocks

[0152]

[0153] 1.1.9 Collision Zone

[0154] A belt of tectonic belts formed by the collision of continents, island arcs, or continents and island arcs. These include suture zones formed by the remnants of ocean basins (ophiolites) representing defunct ancient oceans, and orogenic belts formed by intense compression and contraction, such as the Himalayan collisional orogen and the New Guinea collisional orogen.

[0155] [Annotation example]

[0156] Original text: The Sanshan Island gold mine is located to the east of the subduction collision zone between the Pacific Plate and the Eurasian Plate.

[0157] Triplet:

[0158] Head Entity Type: Deposit

[0159] Head entity: Sanshan Island Gold Mine East

[0160] Relationship: Located

[0161] Tail Entity Type: Collision Band

[0162] Tail entity: subduction collision zone between the Pacific Plate and the Eurasian Plate

[0163]

[0164] 1.1.10 Geographical Location

[0165] Geographical location describes the spatial relationship between an object on Earth's surface (such as a country, region, town, settlement, factory, mine, enterprise, transportation hub, tourist attraction, etc.) and certain external objects. These external objects include natural entities on the Earth's surface and the human conditions formed over history. Spatial relationships include direction and distance. Geographical location varies and is primarily categorized into three types: ① Mathematical geography, which reflects the spatial relationship between an object and the Earth's surface as a whole and is expressed using Earth's longitude and latitude coordinates to describe an object's absolute position; ② Physical geography, which refers to the relative spatial relationship between an object and relevant natural geographical elements (such as mountains, rivers, lakes, seas, deserts, and grasslands); and ③ Political and economic geography, which refers to the spatial relationship between an object and relevant human elements and conditions. For a country, political geography refers to the political conditions and social systems of the surrounding areas and neighboring countries. Internally, it primarily refers to the unique military, political, and administrative system within a larger region. Economic geography refers to the relationship between an object and economic zones, economic centers, settlements, resource production areas, industrial and agricultural bases, transportation routes, and commodity markets. Correctly evaluating the advantages and disadvantages of geographical location will help to give full play to regional advantages, maximize strengths and minimize weaknesses, and rationally arrange industrial layout.

[0166] [Annotation example]

[0167] Original text: The gold exploration area in the northern waters of Sanshan Island is located in the northern waters of Sanshan Island, Laizhou City, Shandong Province. It is located in the northern part of the gold exploration area in the Laizhou-Zhaoyuan area of ​​Shandong Province, one of the first national integrated exploration areas, and within the famous Sanshan Island ore-controlling fault zone in the northwest Jiaozhou area.

[0168] Triplet:

[0169] Head entity type: mining area

[0170] Head entity: Gold exploration area in the northern waters of Sanshan Island

[0171] Relationship: Located

[0172] Tail entity type: Geographic location

[0173] Tail entity: Northern waters of Sanshan Island, Laizhou City, Shandong Province

[0174]

[0175] 1.1.11 Shield

[0176] A region within a craton where Precambrian crystalline basement is exposed over a large area, surrounded by a platform with a caprock, resulting in a shield-like topography. Famous shields worldwide include the Canadian Shield and the Baltic Shield. Because the rock formations exposed in these shields date back to the Archean and Proterozoic eras, their study can yield valuable information about the early history of Earth's evolution. A shield is a relatively elevated unit within a platform. It may have previously been covered by a caprock, though its thickness was significantly lower than that of adjacent areas. The lack of cover on a shield is due to later erosion. Therefore, in terms of tectonic properties, it is essentially the same as a platform.

[0177] [Annotation example]

[0178] Original text: The Nubian Shield is a subduction of the Mozambique Ocean during the Neoproterozoic.

[0179] Triplet:

[0180] Head Entity Type: Shield

[0181] Head Entity: Nubian Shield

[0182] Relationship: Experience

[0183] Tail Entity Type: Construction Event

[0184] Tail Entity: Mozambique Ocean Subduction

[0185]

[0186] 1.1.12 Terrane

[0187] [Annotation example]

[0188] Original text: The Weihai terrane is mainly composed of Neoproterozoic granitic gneiss containing ultrahigh-pressure eclogites.

[0189] Triplet:

[0190] Head entity type: ground body

[0191] Head entity: Weihai terrane

[0192] Relationship: Development

[0193] Tail entity type: rock mass

[0194] Tail body: Neoproterozoic granitic gneiss of UHP eclogite

[0195]

[0196] 1.1.13 Platform

[0197] [Annotation example]

[0198] Original text: Xiao Qinling is located in the east-west tectonic system on the southern edge of the North China Platform and is an important gold producing area in my country. The Xiao Qinling gold deposits are produced in the metamorphic rocks of the Taihua Group of the Neoarchean, and the near east-west secondary detachment fault is the ore-bearing structure.

[0199] Triplet:

[0200] Head entity type: Mineral cluster

[0201] Head entity: Xiao Qinling

[0202] Relationship: Located

[0203] Tail entity type: platform

[0204] Tail entity: North China Platform

[0205]

[0206] 1.1.14 Fault Zone

[0207] Also known as a fault zone. Most fractures in the Earth's crust or lithosphere appear not as a single fracture surface, but rather as multiple fracture surfaces forming a fault zone of a certain width. This belt-like area is either occupied by tectonic rocks produced by faulting, or it can be composed of numerous nearly parallel or intertwined fractures and the rock blocks they separate. A fault zone can sometimes be clearly bounded by major faults developed on either side or one side; at other times, it lacks a clear boundary. Fault zones can range from a few meters to hundreds of meters wide, with large ones reaching tens of kilometers. Fault zone lengths range from meters to kilometers, with the longest reaching thousands of kilometers. The rock layers or rock masses on either side of the fault zone generally experience significant or substantial relative displacement, ranging from a few to tens of centimeters to tens of kilometers, or even cumulatively hundreds of kilometers.

[0208] [Annotation example]

[0209] Original text: Xiadian Gold Mine is located in the middle of the Zhaoping Fault Zone

[0210] Triplet:

[0211] Head Entity Type: Deposit

[0212] Head entity: Xiadian Gold Mine

[0213] Relationship: Located

[0214] Tail entity type: fracture zone

[0215] Tail entity: middle part of Zhaoping fault zone

[0216]

[0217] 1.1.15 Alteration Zone

[0218] In geology, an alteration zone refers to the process in which rocks and minerals, after being exposed to hydrothermal fluids, develop new physical and chemical conditions, resulting in corresponding changes in the structure, texture, and composition of the original rock, and the formation of new mineral assemblages. This change typically does not include the replacement of metallic ore minerals. Because this alteration often occurs in the rocks surrounding the ore bodies of hydrothermal deposits, it is also called wall-rock alteration, and the altered surrounding rock is called altered rock. In short, an alteration zone is an area where the chemical and physical properties of rocks have changed due to hydrothermal fluids.

[0219] [Annotation example]

[0220] Original text: Linglong granite has undergone intense hydrothermal alteration, forming an altered rock belt nearly 100 meters wide.

[0221] Triplet:

[0222] Head entity type: geological phenomenon

[0223] Head entity: hydrothermal alteration

[0224] Relationship: Development

[0225] Tail entity type: geological body

[0226] Tail entity: altered rock belt

[0227]

[0228] 1.2 Mineral belt, including: mineralization belt and mineralization domain.

[0229] 1.3 Mineral clusters and areas include: ore fields, mining (cluster) areas, etc.

[0230] 1.4 Mineral deposits, including gold deposits.

[0231] 1.5 Mineral vein, including: vein body, mineral vein / vein body, etc.

[0232] 1.6 Structure (structure), including faults, folds, ductile shear zones, fractures, etc.

[0233] Geological structures refer to the various structural forms formed in the Earth's crust and its overlying lithosphere during geological processes, including folding, faulting, cracking, and slippage of rock layers. These structures not only record the history of crustal deformation but also reflect the dynamic processes within the Earth. Geological structures are a key area of ​​geological research, crucial for understanding Earth's tectonic movements, the evolution of the Earth's crust, and the distribution and genesis of mineral resources.

[0234] 1.6.1 Fracture

[0235] Also known as fracture. When an object is deformed by force and the stress reaches the breaking strength, it breaks, destroying its continuity and integrity. This is called a fracture or rupture. Fractures in rock include fissures, joints, and faults. Regardless of whether the rock layers or rock masses on either side of the fracture have experienced relative displacement along the fracture surface, fractures can be divided into two basic types based on their mechanical properties: tension fracture and shear fracture.

[0236] [Annotation example]:

[0237] Original text: There are four main groups of faults in the mining area.

[0238] Triplet:

[0239] Head entity type: mining area

[0240] Head entity: mining area

[0241] Relationship: Development

[0242] Tail Entity Type: Broken

[0243] Tail Entity: Broken

[0244]

[0245] 1.6.2 Fault

[0246] A phenomenon in which a rock layer or rock mass fractures under stress, resulting in significant relative displacement along the fracture surface. The fracture surface is called the fault plane, and the intersection of the fault plane and the ground is called the fault line. The rock blocks on either side of the fault plane are called "plates." The plate above the fracture surface is called the "hanging plate," and the plate below is called the "footwall." The distance between the two plates is called the "fault throw." A normal fault in which the hanging plate descends relative to the footwall along the fault plane is called a "normal fault." A reverse fault in which the hanging plate ascends relative to the footwall along the fault plane is called a "reverse fault." A horizontal displacement of the upper and lower plates along the fault plane is called a "thrust fault." Faults are widespread in the Earth's crust and are an important geological structure. The study of faults is of great significance to mineral exploration, mining, engineering and construction, hydrology, water conservancy, as well as volcanology, earthquakes, landforms, and crustal movements.

[0247] [Annotation example]:

[0248] Original text: The north-south normal fault is a late-stage ore-breaking fault

[0249] Triplet:

[0250] Head entity type: Fault

[0251] Head entity: North-South normal fault

[0252] Relationship:

[0253] Tail Entity Type: Broken

[0254] Tail entity: late ore breaking fracture

[0255]

[0256] 1.6.3 Wrinkles

[0257] A collection of folds with a certain degree of continuity is called a fold. The most strongly curved part of a fold is called the hinge, the two wings are called the wings, and the plane dividing the fold into two symmetrical (or equal) parts is called the axial plane. A fold with a horizontal hinge is a flat axial fold; a fold with an inclined hinge is a recumbent fold; and a fold with an upright hinge is a vertical fold. A fold with an upright axial plane is a vertical fold, an inclined axial plane is an inclined fold, and a horizontal axial plane is a recumbent fold. A fold with both wings tilting in the same direction is an inverted fold, and a fold with both wings tilting in the same direction and at the same angle is an isoclinal fold.

[0258] [Annotation example]:

[0259] Original text: Fault and fold structures are widely developed within the gold field.

[0260] Triplet:

[0261] Head entity type: Mineral field

[0262] Head entity: Gold field

[0263] Relationship: Development

[0264] Tail Entity Type: Wrinkle

[0265] Tail entity: fold structure

[0266]

[0267] 1.7 Rock mass, including: ore (minerals), gangue (minerals), ore-bearing host rock, igneous rock, sedimentary rock, metamorphic rock, etc.

[0268] 1.8 Mineralization type, including: deposit genesis type, mineralization, mineralization, etc.

[0269] 1.9 Alteration, including: argillization, advanced argillization, chloritization, epidoteization, sericiteization, carbonatization, feldsparization, pyrite-sericite quartzization, silicification, limoniteization, silicification, pyritization, etc.

[0270] 1.10 Formation age, including: mineralization age, diagenetic age, crystallization age, isotope age, etc.

[0271] 1.10.1 Mineralization Age

[0272] [Annotation example]:

[0273] Original text: The mineralization age of quartz in this area determined by fast neutron activation method of 0Ar-³9Ar is 130~190Ma, which is the early Yanshan period. The mineralization lasted for at least 60Ma.

[0274] Triplet:

[0275] Head entity type: mineralization age

[0276] Head entity: 130~190Ma

[0277] Relationship:

[0278] Tail Entity Type: Construction Event

[0279] Tail entity: Early Yanshanian

[0280]

[0281] 1.10.2 Diagenetic age

[0282] [Annotation example]:

[0283] Original text: The typical Late Jurassic rock body is Linglong granite, the lithology is mainly biotite monzonite, and the diagenetic age is 160~147 Ma.

[0284] Triplet:

[0285] Head entity type: igneous rock

[0286] Head entity: Biotite monzonitic granite

[0287] Relationship:

[0288] Tail entity type: diagenetic age

[0289] Tail body: 160-147 Ma

[0290]

[0291] 1.10.3 Crystallization Age

[0292] [Annotation example]:

[0293] Original text: The crystallization age of Guojialing-type granite is between 123 and 131 Ma.

[0294] Triplet:

[0295] Head entity type: igneous rock

[0296] Head entity: Guojialing type granite

[0297] Relationship:

[0298] Tail Entity Type: Crystallization Age

[0299] Tail body: 123-131 Ma

[0300]

[0301] 1.10.4 Isotopic Age

[0302] Also known as "absolute age," it is the age of a rock's formation calculated based on the content of radioactive decay products in the rock.

[0303] [Annotation example]:

[0304] Original text: Isotopic ages of volcanic rocks in the Laiyang Group are 129.7±1.7Ma, 129.4±2.3Ma, and 131±2Ma.

[0305] Triplet:

[0306] Head entity type: igneous rock

[0307] Head entity: volcanic rocks in the Laiyang Group

[0308] Relationship:

[0309] Tail entity type: isotopic age

[0310] Tail body: 129.7±1.7Ma, 129.4±2.3Ma, and 131±2Ma

[0311]

[0312] 1.11 Geological events, including magmatic activity, tectonic events, etc.

[0313] 1.12 Characteristics, including: geophysical characteristics, geochemical characteristics, ore-controlling structural characteristics, alteration zoning, mineralization zoning, remote sensing image characteristics, isotopic composition, mineral composition, ore structure / texture, mining area characteristics, vein characteristics, rock composition, minerals, grade, homogenization temperature, strike, dip, length, width, thickness, etc.

[0314] 1.12.1 Mineral composition

[0315] Minerals: Natural elements or compounds formed by geological processes. They have a relatively definite chemical composition, and those in solid form also have a fixed internal structure; they are stable within a certain range of physical and chemical conditions and are the basic units that make up rocks and ores. Currently, there are approximately 4,145 known minerals, the vast majority of which are solid inorganic substances. There are only a few liquid minerals (such as natural mercury), gaseous minerals (such as helium), and organic minerals (such as amber). Among solid minerals, the vast majority are crystalline minerals, and only a very few (such as opal) are amorphous minerals. Natural elements or compounds from celestial bodies other than the Earth are called cosmic minerals. Certain elements or compounds that are identical or similar to natural minerals obtained by artificial methods are called synthetic minerals. Natural mineral raw materials and mineral materials are an extremely important type of natural resource, widely used in various sectors of industry, agriculture, and science and technology.

[0316] Mineralogy refers to the basic mineral species and their relative abundance that make up a rock, ore, or geological body. Mineralogy is a crucial aspect of geology, petrology, and mineral deposits, helping us understand the properties, formation processes, and potential economic value of geological bodies.

[0317] [Annotation example]:

[0318] Original text: In the Mesozoic granite exposed in the Linglong gold field, minerals such as biotite, hornblende, and quartz are widely developed.

[0319] Triplet:

[0320] Head entity type: Mineral field

[0321] Head entity: Linglong Gold Field

[0322] Relationship: Development

[0323] Tail entity type: Mineral composition

[0324] Tail entities: biotite, hornblende, quartz

[0325]

[0326] 1.12.2 Geophysical characteristics

[0327] Geophysical anomaly characteristics: a general term for various parameters and elements that describe the intensity, range, shape, scale, zoning, element combination, etc. of geochemical anomalies.

[0328] [Annotation example]:

[0329] Original text: Previous researchers have also carried out a large number of geophysical exploration studies and believed that the unsteady flow of the mantle caused by the westward (present-day orientation) subduction of the ancient Pacific plate beneath the Eurasian continent was the main dynamic background leading to the destruction of the eastern North China craton and the mineralization of "craton-destructive gold deposits".

[0330] Triplet:

[0331] Head Entity Type: Geophysical Features

[0332] Head entity: Unsteady mantle flow caused by the westward (present-day) subduction of the ancient Pacific plate beneath the Eurasian continent

[0333] Relationship: Leading to

[0334] Tail Entity Type: Construction Event

[0335] Tail entity: Destruction of the eastern North China Craton

[0336]

[0337] 1.12.3 Geochemical characteristics

[0338] Geochemistry is the science that studies the chemical composition, distribution, migration, and evolution of elements in the Earth's crust, the Earth, and even the universe. Geochemistry is a frontier discipline that integrates geology and chemistry. It interpenetrates and integrates with other adjacent disciplines (such as marine science, environmental science, astronomy, physics, and biology), giving rise to dozens of sub-disciplines, including marine geochemistry, environmental geochemistry, cosmochemistry, isotope geochemistry, biogeochemistry, and experimental geochemistry. Modern geochemistry has become a flourishing and systematic discipline. Geochemistry has established its own theoretical system, forming a comprehensive disciplinary framework and a comprehensive set of research methods. Along with geology, geophysics, and geodesy, it is one of the four pillars of solid Earth science. Geochemistry not only shoulders the crucial mission of addressing fundamental theoretical questions facing contemporary Earth science—the origin and evolution of celestial bodies, the Earth, life, humanity, and elements—but also provides the basis for providing sufficient mineral resources and a favorable living environment for human society.

[0339] Geochemical anomaly characteristics: a general term for various parameters and elements that describe the intensity, range, shape, scale, zoning, element combination, etc. of geochemical anomalies.

[0340] [Annotation example]:

[0341] Original text: The geochemical characteristics of the vein rocks are characterized by originating from a rich mantle magma source area and may have been modified by the subduction of oceanic crust.

[0342] Triplet:

[0343] Head entity type: Gangue

[0344] Head entity: vein rock

[0345] Relationship: With

[0346] Tail entity type: geochemical characteristics

[0347] Tail entity: originates from the enriched mantle magma source area

[0348]

[0349] 1.12.4 Alteration zoning

[0350] Alteration zoning refers to the phenomenon of zonal changes in the chemical and mineral composition of rocks during geological processes, especially under hydrothermal action. This change reflects the changes in the composition and physical-chemical state of the hydrothermal fluid (such as temperature, H+ concentration, Eh, pressure, etc.) over time or the degree of reaction with the rock. Alteration zoning can be divided into vertical zoning and lateral zoning:

[0351] Vertical zoning: In the vertical direction, the tectonic rock-alteration combination will show changes from shallow to deep. For example, "fracture zone + siliceous veins (clusters), carbonate veins, green mud microveins" may appear in the shallow part, "ductile-brittle deformation + siliceous stockwork, pyrite veinlets" may appear in the middle part, and "ductile deformation + infiltrative silicification + disseminated pyritization, albitization, and ferromagnesian carbonatization" may appear in the deep part.

[0352] Lateral zoning: Horizontally, the dynamic metamorphic zone extends horizontally from the main section or strain center, with deformation-metamorphism decreasing from strong to weak, resulting in tectonic rock (deformation)-alteration zoning. Generally, steeply dipping zones are symmetrical, while gently dipping zones are unidirectional, centered on the main slip section.

[0353] Hydrothermal alteration zoning: When altered rocks react with hydrothermal solutions, their chemical and mineral composition often undergoes zonal changes. This can produce single mineral zones or result in the selective replacement of certain minerals in the altered rock by new mineral phases.

[0354] Alkali metasomatism mineralization model: Alkali metasomatism can be summarized into two major stages: initial alkali introduction and later acid migration, with both silicon and minerals involved in the acid migration. Horizontal alteration zoning on the horizontal plane of the deposit shows alkaline metasomatism in the interior, transitioning to neutral alteration outward, and acidic alteration at the outermost edges. Vertical alteration zoning, or in profile, follows a pattern: alkaline metasomatism always occurs at depth (at the base of the deposit), gradually progressing upward to weakly alkaline and neutral alteration (Ca, Fe, and Mg metasomatism), culminating in acidic and strongly acidic alteration fields.

[0355] The study of alteration zoning is crucial for the exploration and evaluation of mineral deposits because it provides key information such as the location of mineralization centers, depth, and type. By analyzing the assemblage, paragenesis, and zoning characteristics of altered minerals, effective exploration markers can be extracted to guide prospecting efforts.

[0356] [Annotation example]:

[0357] Original text: The common alteration zoning pattern of Jiaodong gold deposits from ore body to surrounding rock is: gold-bearing pyrite quartz veins, pyrite sericite zone, strong-weak pyrite sericite zone, potash granite zone, etc.

[0358] Triplet:

[0359] Head Entity Type: Deposit

[0360] Head entity: Jiaodong Gold Mine

[0361] Relationship: Regularity

[0362] Tail entity type: alteration zoning

[0363] Tail entities: golden pyrite quartz veins, pyrite sericite zone, strong-weak pyrite sericite zone, potash granite zone

[0364]

[0365] 2. The second entity type is used for gold prospecting, including prospecting target areas, remote sensing interpretation anomalies, geophysical anomalies, and geochemical anomalies.

[0366] 3. The third entity type is used to study the genesis of mineral deposits, including: inclusions, material sources, and mineralization conditions.

[0367] 3.1 Inclusions, including fluid inclusions, melt inclusions, etc.

[0368] 3.2 Material sources, including: ore-forming fluids, volatiles, ore-forming metals, etc.;

[0369] 3.3 Mineralization conditions, including temperature, pressure, temperature-pressure, oxygen fugacity, salinity, pH value, redox state, etc.

[0370] 3.3.1 Salinity

[0371] A parameter that indicates the salt content of seawater. The earliest definition of salinity (1889) referred to the total amount of dissolved salts in 1 kg of seawater after evaporation. Later, based on the relationship between chlorinity and salinity, salinity was redefined as the mass (in grams) of inorganic salts contained in 1 kg of seawater after all bromine and iodine in the seawater have been replaced by equivalent amounts of chlorine and all carbonates have been converted to oxides. It is represented by the symbol "S" and is expressed in grams per kilogram. {S}‰ = 0.03 + 1.8050 {chlorinity}‰. In 1978, the concept of "practical salinity" was proposed based on the electrical conductivity of seawater, and K15 was used to define the practical salinity of seawater (S): where K15 is the ratio of the electrical conductivity of a seawater sample to the conductivity of a potassium chloride solution with a mass ratio of 32.4356 × 10-3 at 15°C and a standard atmospheric pressure (101,325 Pa). When K15 is exactly 1, S is exactly 35.

[0372] [Annotation example]:

[0373] Original text: Based on the relationship between the freezing point temperature and salinity of inclusions, Jiao calculated that the salinity of the ore-forming fluids in the Denggezhuang gold deposit is between 1.7% and 16.3% NaCleqv, of which the salinity of type Ib inclusions is between 1.7% and 14.6% NaCleqv, with an average of 8.5% NaCleqv.

[0374] Triplet:

[0375] Head entity type: fluid inclusion

[0376] Head entity: Type I b inclusion

[0377] Relationship: With

[0378] Tail Entity Type: Salinity

[0379] Tail body: 1.7% to 14.6% NaCleqv, average 8.5% NaCleqv

[0380]

[0381] 3.3.2 pH value

[0382] The negative logarithm of [H+] is commonly used to compare the acidity or alkalinity of a solution. This negative logarithm is called pH. Its calculation is as follows: pH = -1 g [H+]. The relationship between the acidity or alkalinity of a solution and its pH is: a neutral solution has a pH of 7, an acidic solution has a pH < 7 (the lower the pH, the stronger the acidity), and an alkaline solution has a pH > 7 (the higher the pH, the stronger the alkalinity). pH is one of the important factors affecting element migration and precipitation.

[0383] [Annotation example]:

[0384] Original text: Coke can change the pH value of the ore-bearing fluid during the sericite alteration process, and the pH value of the ore-bearing fluid in turn affects the solubility of metal minerals. , and increase the pH value of the solution to make it weakly alkaline.

[0385] Triplet:

[0386] Head Entity Type: Magmatic Activity

[0387] Head entity: reaction process between fluid and surrounding rock

[0388] Relationship: With

[0389] Tail entity type: pH

[0390] Tail entity: The solution pH value increases and becomes weakly alkaline.

[0391]

[0392] 2. Relationship Type

[0393] Entities of relationship type can be as follows: include, for, use, have, cause, form, lead to, summarize, speculate / predict, according to, think, explain, only / only, exceed, bounded by, centered on, between (among), after, later than, earlier than, formed in, located in, concurrent with, take... as an example, concentrate on, related to, benefit from, provide, produce in, output, divide into, distribute in, controlled by, control, superimpose on, spread / expand along..., extend, meet, surround, surround, fill, host, enrich, symbiotic, develop, migrate, intrude, alteration, associated, host, display, prospecting target, mineralization.

[0394] Furthermore, when extracting triple data, the geological prospecting knowledge triples of the target file can be extracted step by step using the prompt word engineering in combination with the prospecting knowledge graph model. Figure 4 The following process can be streamlined using the Agent concept to achieve automatic triple extraction.

[0395] In the embodiment of the present invention, each step (i.e., Step 1 to Step 6) can be implemented based on the prompt word engineering combined with the prospecting knowledge graph ontology model.

[0396] Step 1:

[0397] You are a professional gold prospector and gold deposit scientist. I will input a paragraph from a document. Your task is to extract triples related to gold prospecting from this paragraph as comprehensively as possible.

[0398] Task execution steps:

[0399] 1) Sentence-by-sentence analysis: First, the document paragraph is split sentence by sentence. Independent triples are extracted for each sentence to ensure that no details are missed.

[0400] 2) Triple Extraction: For each sentence, triples are extracted using the following steps: - Identify the head entity: Find the primary object or concept discussed in the sentence. - Identify the relationship: Analyze the relationship between the head entity and other components in the sentence. - Locate the tail entity: Find the object or concept connected to the head entity through a relationship.

[0401] 3) Record the original text: Attach the original sentence from the original text to each extracted triple for subsequent verification and screening.

[0402] 4) Considering Inter-Sentence Relationships: After completing sentence-by-sentence extraction, review the logical relationships between sentences to check for any triplets that can be extracted across sentences. Extraction Guidelines: - Relationship descriptions should be specific and close to the original text, avoiding over-interpretation. - Ignore formatting and filtering; these will be addressed later. - Emphasize key information such as numerical values, names of people, places, mining areas, geological bodies, and time. - Each sentence should be carefully reviewed to extract all potential triplets. - Potentially duplicate triplets should be extracted, and filtering will be performed after extraction is complete.

[0403] Output format example: (head entity, relationship, tail entity): Original text paragraph: The Jiaodong gold mining area covers approximately 16,522 square kilometers, accounting for only 0.17% of China's landmass. Its cumulative proven gold reserves and gold production account for approximately one-quarter of China's total, making it China's most important gold base. Gold exploration and research in this region are a hot topic of great interest both domestically and internationally, with continued progress in prospecting. Before 2005, over 1,700 tons of gold reserves were discovered at depths of less than 500 meters; since 2005, over 2,700 tons have been discovered at depths between 500 and 2,000 meters. Currently, the deepest gold exploration drill hole in the Jiaodong region reaches a depth of 4,006.17 meters. New insights have been continuously gained from the research on the mineralization laws and prospecting theories of the Jiaodong gold deposits. For example, the Jiaojia-type gold deposit type discovered early on in the altered rocks of the fracture zone broke the traditional understanding that large faults only guide mineralization but do not form minerals; the stepped mineralization model proposed recently in deep prospecting has solved the problems of deep ore-bearing location and prospecting direction; the Jiaodong gold deposits were previously believed to be greenstone belt-type gold deposits or orogenic gold deposits, but now the thermal uplift-extension mineralization theory has been proposed; predecessors believed that the gold deposits were formed in the Proterozoic or multi-stage mineralization, but now it is generally believed that the gold deposits were formed in the Cretaceous.

[0404] Please output the extracted triples.

[0405] Step 2

[0406] You are a professional geologist and gold mineralogist. After completing the initial triple extraction, you need to further address the issue of ambiguous entity references to ensure that the entities in each triple accurately correspond to specific objects. The following is a detailed guide:

[0407] Task execution steps:

[0408] 1) Identify ambiguous entities: Review the initially extracted triples and identify all entities that may have ambiguous references.

[0409] 2) Contextual analysis: For each unidentified entity, we go back to the original text and analyze the context in which it appears to determine its exact reference.

[0410] 3) Entity Association: Based on contextual information, associate ambiguous entities with entities explicitly mentioned in the text. Specific guidelines are as follows: - If the entity appears for the first time in a paragraph and is clearly defined, associate it directly. - If the entity is mentioned using a pronoun or an abbreviation, find the original entity it refers to and replace or supplement it in the triple. - For entities mentioned consecutively, ensure consistency of reference.

[0411] 4) Update triples: Update triples based on the clarified entity references to ensure clarity and specificity of each entity. Example: 1. Original sentence: "Gold deposits are primarily located within a thick tectonic alteration zone between the Muniushan intrusion, the Jingshan Group, and the Queshan intrusion." - Clarified entity: The gold deposits should refer to the gold deposits of the Qianchuiliu Gold Mine. 2. Original sentence: "Gold-bearing pyrites from typical ore bodies were selected as the research objects." - Clarified entity: The typical ore bodies should refer to the typical ore bodies of the Qianchuiliu Gold Mine. 3. Original sentence: "The S and Pb isotopic characteristics of the deposit indicate a mixed crust-mantle source of ore-forming materials." - Clarified entity: The deposit should refer to the Qianchuiliu Gold Mine. 4. Original sentence: "This reflects a large-scale Yanshanian mineralization event in the area, indicating significant prospecting potential in the northeastern margin of the Jiaolai Basin." - Clarified entity: The area should refer to the northeastern margin of the Jiaolai Basin in Jiaodong. Output format example: (clarified head entity, relation, clarified tail entity): original sentence.

[0412] Please review and update the initially extracted triples according to the above steps to ensure that the reference of each entity is clear and accurate. This will help improve the accuracy and usability of our dataset.

[0413] Step 3

[0414] You are a professional geological prospector and gold deposit scientist. After completing entity clarification, you need to further determine the type of the head and tail entities in the extracted triples. The following is a detailed operation guide:

[0415] Steps:

[0416] 1) Entity non-empty verification: First, check whether the head entity and tail entity in each triple are empty. If either entity is empty, delete the triple.

[0417] 2) Entity type determination: After confirming that the entity is not empty, determine the entity type. Please classify the head entity and tail entity according to the following entity type list. If the entity type does not belong to the following list, delete the triple: - The entity type list includes: "Head entity detailed type" and "Tail entity detailed type" are limited to the following detailed classifications:

[0418] Plateau, basin, craton, plate, rift, subduction zone, suture zone, orogenic belt, collision zone area, geographical location, shield, terrane, platform, fault zone, alteration zone, metallogenic belt, metallogenic domain, ore cluster, ore field, mining area, ore section, ore deposit, vein, vein body, stratum, fault, fold, ductile shear zone, fracture, ore mineral, ore, gangue, gangue mineral, ore-bearing wall rock, igneous rock, sedimentary rock, metamorphic rock, ore deposit genesis type, mineralization, mineralization, wall rock alteration type, mudification, advanced mudification, chloritization, epidote, sericitization, carbonatization, potassium feldsparization, pyrite sericitization, silicification, limonite mineralization, mineralization age, diagenetic age, crystallization age, Isotope age, magmatic activity, tectonic events, mineralization mechanism, mineralization stage, reserves, geophysical characteristics, geochemical characteristics, ore-controlling structural characteristics, alteration zoning, mineralization zoning, remote sensing image characteristics, isotope composition, mineral composition, ore texture, ore structure, mining area characteristics, vein characteristics, rock composition, minerals, grade, homogenization temperature, strike, dip, length, width, thickness, prospecting target area, favorable mineralization location, mineralized enrichment zone, remote sensing interpretation anomalies, geophysical anomalies, geochemical anomalies, fluid inclusions, melt inclusions, ore-forming fluids, volatiles, ore-forming metals, temperature, pressure, temperature-pressure, oxygen fugacity, salinity, pH value, redox state.

[0419] Output format example: (head entity, relation, tail entity): original sentence, head entity type, tail entity type. Please strictly follow the above steps to ensure the accuracy and completeness of the triple dataset.

[0420] Step 4

[0421] You are a professional geological prospector and gold deposit scientist. After determining the type of the head and tail entities in the triple, you need to summarize the types. The following is a detailed operation guide:

[0422] Steps:

[0423] Read the list of entity types: The list of entity types includes: plateau, basin, craton, plate, rift, subduction zone, suture zone, orogenic belt, collision zone area, geographical location, shield, terrane, platform, fault zone, alteration zone, mineralization belt, mineralization domain, ore cluster, ore field, mining area, ore section, ore deposit, vein, vein body, stratum, fault, fold, ductile shear zone, fracture, ore mineral, ore, gangue, gangue mineral, ore-bearing host rock, igneous rock, sedimentary rock, metamorphic rock, ore deposit genesis type, mineralization, mineralization, host rock alteration type, argillization, advanced argillization, chloritization, epidoteization, sericiteization, carbonatization, potassium feldsparization, pyrite-sericite quartzization, silicification, limonite mineralization, mineralization age, Diagenetic age, crystallization age, isotopic age, magmatic activity, tectonic events, mineralization mechanism, mineralization stage, reserves, geophysical characteristics, geochemical characteristics, ore-controlling structural characteristics, alteration zoning, mineralization zoning, remote sensing image characteristics, isotopic composition, mineral composition, ore texture, ore structure, mining area characteristics, vein characteristics, rock composition, minerals, grade, homogenization temperature, strike, dip, length, width, thickness, prospecting target area, favorable mineralization location, mineralized enrichment zone, remote sensing interpretation anomalies, geophysical anomalies, geochemical anomalies, fluid inclusions, melt inclusions, ore-forming fluids, volatiles, ore-forming metals, temperature, pressure, temperature-pressure, oxygen fugacity, salinity, pH value, redox state.

[0424] The "head entity summary type" and "tail entity summary type" are determined by the entity type list. For example, "region" should include plateaus, basins, cratons, plates, rifts, subduction zones, suture zones, orogenic belts, collision zones, geographical locations, shields, terranes, platforms, fault zones, alteration zones, etc.; "mineral clusters" should include ore fields, mining areas, mineral clusters, etc.; "mineral belts" should include mineralization belts, mineralization domains, etc.; "mineral sections" should include mineral sections; "mineral deposits" should include gold mines (deposits), etc. All "head entity summary types" and "tail entity summary types" are limited to the contents in the "head entity detailed type" and "tail entity detailed type" descriptions.

[0425] Output format example: (head entity, relationship, tail entity): original sentence, head entity summary type, head entity type, tail entity summary type, tail entity type. Please strictly follow the above steps to ensure the accuracy and completeness of the triple dataset.

[0426] Step 5

[0427] You are a professional geological prospector and gold deposit scientist. After completing entity non-null verification and type determination, you need to further screen and determine the relationships between entities in the triple. The following is a detailed operation guide:

[0428] Steps:

[0429] 1) Relationship screening: Check whether the relationship in each triple can be mapped to one of the following given relationship lists.

[0430] 2) Relationship determination: Based on the original text content, determine whether the relationship in the triple can be accurately converted into a relationship in the relationship list.

[0431] If a relationship is not in the given relationship list, or cannot be accurately mapped to a relationship in the list based on the original text, the triplet is marked as invalid and deleted. The given relationship list includes: belong to, located in, developed, experienced, related, controlled, displayed, predicted, prospecting target, associated, superimposed, formed in, similar, consistent, inconsistent, superimposed on, stronger than, weaker than, have, contain, developed, overlying, underlying, including, divided into, range, distributed, for, used for, genesis, formed, led to, summarize, speculate / predict, according to..., believe, explain, only / only, beyond, bounded by..., centered on..., between (in)..., after, later than, earlier than, concurrent with..., taking... as an example, concentrated on, beneficial to, provide, produced in, output, distributed in, controlled by, spread along..., extend, meet, surround, surround, fill, host, enrich, symbiotic, migrate, intrude, alteration, host, display, mineralization.

[0432] Output format example: (head entity, relationship, tail entity): original sentence, head entity summary type, head entity type, tail entity summary type, tail entity type, relationship type. Please perform relationship determination according to the above guidelines and compile a list of triples that meet the requirements.

[0433] Step 6

[0434] You are a professional geological prospector and gold deposit scientist. Now we need to convert the confirmed triples into a table in a specified format. The following are the specific operation requirements: Operation requirements:

[0435] 1) Create a table: Construct the table using the following column headings: - Sequence number - Header entity summary type - Header entity type - Header entity - Relationship between entities - Tail entity summary type - Tail entity type - Tail entity - Relationship attributes - Other attributes - Original text - Article number - Remarks.

[0436] 2) Fill in the following information: - Sequence Number: Assign a unique number to each triple in sequence. - Head Entity Type, Head Entity, Relationship Between Entities, Tail Entity Type, Tail Entity: Fill in the fields based on the results of the previous step. - Relationship Attribute: Based on the original text, label it as "Opinion," "Fact," or "Method." - Other Attributes: Based on the entity content, label it as "Value," "Person," "Place," "Time," or a combination of these attributes. - Original Sentence: Quote directly from the original text. - Article Number: Enter "roy." - Remarks: Leave this blank for future reference. Important Notes: - Ensure the information is accurate and formatted consistently. - Carefully label relationship attributes and other attributes to reflect the precise meaning of the original text. Please format the table according to the above requirements.

[0437] Step 7

[0438] You are a professional geological prospector and gold mineralogist. Now you need to remove duplicates from the formatted triple table to ensure the conciseness of the results.

[0439] Target:

[0440] Remove duplicate triples to avoid triplets with the same or similar meanings from appearing in the final results.

[0441] Make sure each triplet is relevant to gold prospecting.

[0442] Ensure that the format and content of each triplet meet the requirements. Please start deduplication.

[0443] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiment. All technical solutions based on the concept of the present invention are within the scope of protection of the present invention. It should be noted that for those skilled in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for extracting geological prospecting knowledge triples using prompt word engineering, characterized in that: The method comprises: Defining a prospecting knowledge graph ontology model; the prospecting knowledge graph ontology model includes multiple entity types and multiple relationship types; Combined with the prospecting knowledge graph model, the prompt word project is used to extract the geological prospecting knowledge triples of the target file step by step; specifically, the method includes: parsing the target file into a text in an article format and segmenting the text into paragraphs; combined with the prospecting knowledge graph model, the prompt word project is used to extract the geological prospecting knowledge triples of each paragraph step by step; and the triple extraction results corresponding to each paragraph are merged into a table; Among them, combined with the prospecting knowledge graph model, the prompt word engineering is used to extract the geological prospecting knowledge triples of each paragraph step by step, including: Extract the triples in the current paragraph; Processing the triples to resolve the issue of unclear entity references in the triples, specifically including: identifying unclear references: reviewing the initially extracted triples to identify entities with unclear references; context analysis: for each unclear entity, going back to the original paragraph and analyzing the context in which it appears to determine its exact reference; entity association: associating the unclear entity with the entity clearly mentioned in the text based on the context information; updating triples: updating the triples based on the clarified entity references to ensure that each entity is clear and unambiguous; Determine and filter the head entity type and the tail entity type; summarize the head entity type and the tail entity type; determine and filter the relationship between the head entity and the tail entity; format the triple list into a table; deduplicate the table; and save this output as the current paragraph extraction result.

2. The method according to claim 1, characterized in that The entity types include a first entity type for basic geological information, a second entity type for gold prospecting, and a third entity type for studying the genesis of ore deposits; The first entity type includes regional geological background or mineral deposit geological background, including region, mineral belt, mineral cluster, mineral deposit, mineral section, vein, stratum, structure, rock mass, mineralization type, alteration, formation age, geological event, mineralization mechanism, mineralization stage, reserves and other characteristics; the other characteristics include geophysical characteristics, geochemical characteristics, ore-controlling structural characteristics, alteration zoning, mineralization zoning, remote sensing image characteristics, isotopic composition, mineral composition, ore structure / structure, mining area characteristics, vein characteristics, rock composition, minerals, grade, homogenization temperature, strike, dip, length, width and thickness; The second entity type includes prospecting target areas, remote sensing interpretation anomalies, geophysical anomalies, and geochemical anomalies; The third entity type includes: inclusions, material sources, and mineralization conditions.

3. The method according to claim 1, characterized in that Extracting triples from the current paragraph content includes: Sentence-by-sentence analysis: The document paragraph is broken down sentence by sentence, and independent triples are extracted for each sentence. This includes: identifying the head entity: finding the main object or concept discussed in the sentence; determining the relationship: analyzing the relationship between the head entity and other components in the sentence; and locating the tail entity: finding another object or concept connected to the head entity through a relationship. Record the original text: Attach the original sentence from the original text to each extracted triple for subsequent verification and screening; Consideration of inter-sentence relations: After completing sentence-by-sentence extraction, review the logical relations between sentences and check whether there are triplets across sentences that can be extracted.

4. The method according to claim 1, wherein Determining and filtering the head entity type and tail entity type includes: Entity non-empty verification: Check whether the head entity and tail entity in each triple are empty. If either entity is empty, delete the corresponding triple; Entity type determination: After confirming that the entity is not empty, the entity type is determined; the head entity and the tail entity are classified according to the preset entity type list. If the entity type does not belong to any type in the entity type list, the corresponding triple is deleted.

5. The method according to claim 1, wherein Determining and filtering the relationship between the head entity and the tail entity includes: Relationship screening: Check whether the relationship in each triple can be mapped to one of the preset relationship lists; Relationship determination: Based on the original content, determine whether the relationship in the triple can be accurately converted to any relationship in the relationship list; if the relationship is not in the preset relationship list, or cannot be accurately mapped to any relationship in the relationship list based on the original content, mark it as invalid and delete the triple.

6. The method according to claim 1, characterized in that Formatting a list of triplets into a table involves: Create a table: Use the following column headings to construct the table: - Sequence number - Head entity summary type - Head entity type - Head entity - Relationship between entities - Tail entity summary type - Tail entity type - Tail entity - Relationship attributes - Other attributes - Original text - Article number - Remarks; Fill in the form information.

7. A device for extracting geological prospecting knowledge triples using prompt word engineering, characterized in that: It includes a processor and a memory storing program instructions, and is characterized in that the processor is configured to execute the method of extracting geological prospecting knowledge triples using prompt word engineering as described in any one of claims 1 to 6 when executing the program instructions.

Citation Information

Patent Citations

  • Contract risk identification method and device, equipment, storage medium and program product

    CN119205338A