Mechanical knowledge graph construction method and device, equipment and storage medium

Abstract

The invention discloses a mechanical knowledge graph construction method and device, equipment and a storage medium, and the method comprises the steps: carrying out the entity and relation extraction of each mechanical knowledge document based on a large language model, and enabling the mechanical knowledge document to comprise a maintenance work order, an overhaul manual and a product manual; constructing an entity relationship sub-graph according to entities and relationships extracted from the mechanical knowledge documents; and connecting the entity relationship sub-graphs by adopting a graph construction algorithm to obtain the mechanical knowledge graph. For an engineering machinery knowledge management scene, entity and relationship extraction is performed on diversified data in the machinery field based on a large language model, and a machinery knowledge graph is automatically constructed based on the extracted entities and relationships based on a graph construction algorithm, so that the extraction efficiency and accuracy of the entities and the relationships are improved, and the engineering machinery knowledge management efficiency is improved. Therefore, efficient and stable construction of the mechanical knowledge graph is realized.

Description

Technical Field

[0001] The present invention relates to the field of mechanical technology, and in particular to a method, device, equipment and storage medium for constructing a mechanical knowledge graph. Background Art

[0002] With the rapid development of the construction machinery industry, digital management of knowledge has become a key link in improving the competitiveness of enterprises. Within the existing technical framework, knowledge management in the field of construction machinery mainly relies on traditional databases and document systems. However, knowledge graphs are gradually becoming the mainstream choice to replace traditional data carriers due to their intuitive expressiveness, efficient query capabilities and the diversity of knowledge. At the same time, thanks to their flexible relational structure, knowledge graphs provide high-quality reasoning basis for large models, laying an important foundation for the digital transformation and intelligent upgrading of enterprises.

[0003] However, in the field of engineering machinery, the construction process of knowledge graphs mainly relies on single entity extraction, expert experience and manual screening, which not only has low accuracy but also complex processes. In the context of the current explosive growth of data volume and the diversification of engineering machinery knowledge, it is difficult to achieve efficient and stable construction of knowledge graphs. Summary of the invention

[0004] The present invention provides a method for constructing a mechanical knowledge graph to achieve efficient and accurate construction of a mechanical knowledge graph.

[0005] According to a first aspect of the present invention, a method for constructing a mechanical knowledge graph is provided, comprising: extracting entities and relationships from each mechanical knowledge document based on a large language model, wherein the mechanical knowledge document includes a maintenance work order, an overhaul manual, and a product manual;

[0006] constructing an entity-relationship subgraph according to the entities and relationships extracted from each of the mechanical knowledge documents;

[0007] A graph construction algorithm is used to connect the entity relationship subgraphs to obtain a mechanical knowledge graph.

[0008] According to another aspect of the present invention, a mechanical knowledge graph construction device is provided, comprising: an entity and relationship extraction module, for extracting entities and relationships from each mechanical knowledge document based on a large language model, wherein the mechanical knowledge document includes a maintenance work order, an overhaul manual, and a product manual;

[0009] An entity-relationship subgraph construction module, used for constructing an entity-relationship subgraph according to entities and relationships extracted from each of the mechanical knowledge documents;

[0010] The mechanical knowledge graph acquisition module is used to connect the entity relationship subgraphs by adopting a graph construction algorithm to obtain a mechanical knowledge graph.

[0011] Optionally, entities and relations are extracted from each mechanical knowledge document based on a large language model, including:

[0012] Extracting entities from each of the mechanical knowledge documents based on a large language model according to a preset graph architecture, and deleting the extracted identical entities using an entity disambiguation method, wherein the entities include product line names, system names, component names, parameters, and fault types;

[0013] Relationships corresponding to the entities are extracted from the mechanical knowledge document, wherein the relationships include assembly relationships between components, failures occurring in components, and relationships between components and parameters.

[0014] Optionally, constructing an entity-relationship subgraph according to the entities and relationships extracted from each of the mechanical knowledge documents includes:

[0015] Acquire a triple determined by entities and relations extracted from each mechanical knowledge document, wherein the triple includes a relation and a pair of entities;

[0016] evaluating the relation confidences in the triples, and screening the triples according to the relation confidences;

[0017] The entity relationship subgraph corresponding to each of the mechanical knowledge documents is constructed according to the filtered triples.

[0018] Optionally, a graph construction algorithm is used to connect the entity relationship subgraphs to obtain a mechanical knowledge graph, including:

[0019] Acquire an expert knowledge tree, wherein the expert knowledge tree includes standard entities and relationships in the mechanical field;

[0020] The graph construction algorithm is used to connect each of the entity relationship subgraphs based on the expert knowledge tree to obtain a mechanical knowledge graph.

[0021] Optionally, the graph construction algorithm is used to connect the entity relationship subgraphs based on the expert knowledge graph to obtain a mechanical knowledge graph, including:

[0022] Determine common entities in the expert knowledge tree and each of the entity relationship subgraphs, wherein the common entities include the same entity or similar entities;

[0023] The common entity is used as an alignment node, and based on the alignment node, the expert knowledge tree and each entity relationship subgraph are connected to obtain the mechanical knowledge graph.

[0024] Optionally, determining common entities in the expert knowledge tree and each of the entity relationship subgraphs includes:

[0025] Calculating the similarity between the expert knowledge tree and each entity in the entity relationship subgraph;

[0026] The two entities whose similarity is equal to the first value are regarded as the same entity, and the two entities whose similarity is greater than the second value and less than the first value are regarded as the similar entities.

[0027] Optionally, after the graph construction algorithm is used to connect the entity relationship subgraphs to obtain the mechanical knowledge graph, the method further includes:

[0028] Obtaining a mechanical maintenance request input by a user, and extracting a target entity contained in the mechanical maintenance request;

[0029] The mechanical knowledge graph is queried based on the target entity to obtain associated entities corresponding to the target entity, and maintenance suggestions are generated based on the associated entities.

[0030] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0031] at least one processor; and

[0032] a memory communicatively connected to the at least one processor; wherein,

[0033] The memory stores a computer program that can be executed by the at least one processor. The computer program is executed by the at least one processor so that the at least one processor can perform the method described in any embodiment of the present invention.

[0034] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the method described in any embodiment of the present invention when executed.

[0035] The technical solution of the embodiment of the present invention is aimed at the construction machinery knowledge management scenario. It extracts entities and relationships from diversified data in the machinery field based on a large language model, and automatically constructs a machinery knowledge graph based on the extracted entities and relationships based on a graph construction algorithm, thereby improving the efficiency and accuracy of entity and relationship extraction, thereby realizing efficient and stable construction of the machinery knowledge graph.

[0036] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0038] Figure 1 is a flowchart of a method for constructing a mechanical knowledge graph according to the first embodiment of the present invention;

[0039] Figure 2 is a schematic diagram of a graph architecture provided according to Embodiment 1 of the present invention;

[0040] Figure 3 is a flowchart of a method for constructing a mechanical knowledge graph according to the second embodiment of the present invention;

[0041] Figure 4 is a schematic diagram of the structure of a mechanical knowledge graph construction device provided according to the third embodiment of the present invention;

[0042] Figure 5 It is a schematic diagram of the structure of an electronic device provided by Embodiment 4 of the present invention. DETAILED DESCRIPTION

[0043] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme 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 only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0044] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0045] Embodiment 1

[0046] Figure 1A flowchart of a model detection method is provided for the first embodiment of the present invention. This embodiment is applicable to the case where a model is detected based on a model detection request. The method can be executed by a model detection device, which can be implemented in the form of hardware and / or software. Figure 1 As shown, the method includes:

[0047] Step S101 , extracting entities and relations from each mechanical knowledge document based on a large language model.

[0048] Optionally, entities and relationships are extracted from each mechanical knowledge document based on a large language model, including: extracting entities from each mechanical knowledge document based on a preset graph architecture based on the large language model, and deleting the extracted identical entities using an entity disambiguation method, wherein the entities include product line names, system names, component names, parameters, and fault types; extracting relationships corresponding to the entities from the mechanical knowledge documents, wherein the relationships include assembly relationships between components, faults occurring in components, and relationships between components and parameters.

[0049] Specifically, the mechanical knowledge documents in this embodiment include maintenance work orders, overhaul manuals and product manuals. Of course, this embodiment is only an example and does not limit the specific content of the mechanical knowledge documents. In addition, a knowledge graph extraction tool, such as DeepKE-LLM, can be used in this embodiment. This embodiment does not limit the specific type of the knowledge graph extraction tool, and a large language model that applies natural language processing technology is also integrated in the knowledge graph extraction tool. Therefore, the knowledge graph extraction tool specifically performs entity recognition on the input mechanical knowledge document based on the large language model, and extracts entities with specific meanings in the document. At the same time, it can also identify the semantic relationships between entities in the document, so as to build a complex connection network between entities.

[0050] Among them, in this implementation, the large language model with fixed value fine-tuning is adapted to the knowledge graph extraction tool, and when performing entity extraction, it will refer to the following Figure 2 The graph architecture shown extracts entities from various mechanical knowledge documents. For example, entities include product line names, system names, component names, parameters, and fault types. Of course, this embodiment is only an example and does not limit the specific types of entities extracted. In addition, the large language model performs entity extraction step by step, and each time the extracted entities are processed by entity disambiguation to avoid overlap with existing entities. When performing entity disambiguation, the embedding vectors corresponding to the currently extracted entity and the existing entities are first calculated, and the following formula (1) is used to determine the similarity value between the currently extracted entity embedding vector and the existing entity embedding vector:

[0051] sim(e,e′)=e·e′ / |e|·|e′| (1)

[0052] Among them, e and e' are the embedding vectors of two entities, and the similarity value of the embedding vector is used as the similarity value of the entity. If the similarity value between the currently extracted entity and any existing entity exceeds the preset threshold, it means that the same entity as the currently extracted entity already exists, so the currently extracted entity can be deleted to avoid duplication of entity extraction. In this embodiment, the entity disambiguation method can avoid extracting the same entity for the same mechanical knowledge document.

[0053] In addition, in this embodiment, after the entity is extracted, the relationship corresponding to the entity will also be extracted from the mechanical knowledge document, such as the assembly relationship between components, the failure of components, and the relationship between components and parameters. Of course, this embodiment is only an example for illustration and does not limit the specific type of extracted relationship.

[0054] Step S102: construct an entity relationship subgraph according to the entities and relationships extracted from each mechanical knowledge document.

[0055] Optionally, an entity relationship subgraph is constructed based on the entities and relationships extracted from each mechanical knowledge document, including: obtaining triples determined by the entities and relationships extracted from each mechanical knowledge document, wherein the triples include a relationship and a pair of entities; evaluating the relationship confidence in the triples, and filtering the triples according to the relationship confidence; and constructing an entity relationship subgraph corresponding to each mechanical knowledge document based on the filtered triples.

[0056] Optionally, evaluating the confidence of the relationship in the triples includes: calculating the frequency of occurrence of each relationship in the triples extracted from each mechanical knowledge document; and determining the confidence of the relationship according to the frequency of occurrence.

[0057] Specifically, in this implementation, multiple triples will be determined for the entities and relationships extracted from each mechanical knowledge document. Specifically, one triple will be determined for each relationship and a pair of entities, and an entity relationship subgraph corresponding to each mechanical knowledge document can be constructed based on the determined multiple triples. Among them, before constructing the entity relationship subgraph based on the triples, the relationship confidence in the triples will be evaluated first. When evaluating the relationship confidence, the frequency of occurrence of each relationship in the triples extracted from each mechanical knowledge document is calculated, and the frequency of occurrence of the relationship is used as the relationship confidence. For example, relationship X appears 10 times, and 10 is used as the confidence of relationship X. When the value of the relationship confidence is relatively large, it means that the importance of this relationship between entities is relatively high. At this time, the three groups with relationship confidence values ​​exceeding the specified value will be retained, and the ones with relatively low relationship confidence values ​​will be screened out. For example, 100 triples are extracted for a mechanical knowledge document through a large language model, but only 80 are left after relationship confidence screening. At this time, the entity relationship subgraph corresponding to the mechanical knowledge document will be constructed based on the remaining 80 triples after screening.

[0058] It should be noted that, since the triples extracted for each mechanical knowledge document may be interrupted, the mechanical knowledge document corresponds to a set of entity relationship subgraphs, such as two entity relationship subgraphs. In this embodiment, the specific number of entity relationship subgraphs extracted from each mechanical knowledge document is not limited, but a set of entity relationship subgraphs extracted from a mechanical knowledge document does not contain the same entity. Since the number of mechanical knowledge documents input can be multiple, the number of corresponding entity relationship subgraphs obtained is also relatively large, and is not less than the number of mechanical knowledge documents.

[0059] Step S103: Use a graph construction algorithm to connect the entity relationship subgraphs to obtain a mechanical knowledge graph.

[0060] Optionally, a graph construction algorithm is used to connect each entity relationship subgraph to obtain a mechanical knowledge graph, including: obtaining an expert knowledge tree, wherein the expert knowledge tree includes standard entities and relationships in the mechanical field; and using a graph construction algorithm to connect each entity relationship subgraph based on the expert knowledge tree to obtain a mechanical knowledge graph.

[0061] Optionally, a graph construction algorithm is used to connect each entity relationship subgraph based on the expert knowledge graph to obtain a mechanical knowledge graph, including: determining common entities in the expert knowledge tree and each entity relationship subgraph, wherein the common entities include the same entity or similar entities; using the common entities as alignment nodes, and connecting the expert knowledge tree and each entity relationship subgraph based on the alignment nodes to obtain a mechanical knowledge graph.

[0062] Optionally, determining common entities in the expert knowledge tree and each entity relationship subgraph includes: calculating the similarity between entities in the expert knowledge tree and each entity relationship subgraph; treating two entities whose similarity is equal to a first value as the same entity, and treating two entities whose similarity is greater than a second value and less than the first value as similar entities.

[0063] Specifically, in this embodiment, a graph construction algorithm is used to combine the expert knowledge tree and the obtained entity relationship subgraph to construct a mechanical knowledge graph, wherein the expert knowledge tree includes standard entities and relationships in the mechanical field. By supplementing the entity relationship subgraph with the expert knowledge tree, the professionalism and accuracy of the mechanical knowledge graph can be ensured. In this embodiment, the following formula (2) is used to determine the entity similarity relationship in the expert knowledge tree and each entity relationship subgraph:

[0064]

[0065] Among them, R(A, B) represents the similarity relationship between entity A and entity B. When the similarity is α, the two are the same entity; when the similarity is greater than β, the two are similar entities; when the similarity is less than or equal to β, the two are unrelated entities. In this embodiment, the same entity and similar entities can be regarded as common entities.

[0066] It should be noted that in this embodiment, after obtaining the common entities, the common entities are used as alignment nodes, and based on the alignment nodes, the expert knowledge tree and each entity relationship subgraph are connected to obtain the mechanical knowledge graph. Figure 1 contains entity A, and in the entity relationship sub Figure 2 If entity A is also included in the entity relationship Figure 1 and entity relationship sub Figure 2 According to entity A, the entities are aligned and connected. Of course, this embodiment only takes the example of connecting the entity relationship subgraph based on an alignment node as an example, but in actual applications, multiple common entities can be determined, and multiple entity relationship subgraphs and an expert tree can be connected together based on multiple common entities to obtain the mechanical knowledge graph. Among them, since the expert knowledge tree includes standard entities and relationships in the mechanical field, and the number is large, it can ensure that the mechanical knowledge graph is continuous and uninterrupted, and ensures professionalism and accuracy.

[0067] It is worth mentioning that in this embodiment, when the common entity is a similar entity, for example, entity A and entity B are similar entities, then in addition to aligning the entity relationship subgraph based on the similar entities, this application will also store the name of entity B in the alias attribute of entity A, thereby ensuring the comprehensiveness of the acquired mechanical knowledge graph. Therefore, through entity alignment, relationship linking and graph optimization, the algorithm combines the entity relationship subgraphs into a complete mechanical knowledge graph. Therefore, in this embodiment, in response to the scenario requirements of construction machinery, more accurate entity and relationship extraction is achieved through customized graph architecture and relationship confidence constraints. The use of expert knowledge trees to integrate domain knowledge improves the professionalism of the knowledge graph. At the same time, the graph construction algorithm ensures the richness and accuracy of the graph.

[0068] In the implementation mode of the present application, for the engineering machinery knowledge management scenario, entities and relationships are extracted from diversified data in the machinery field based on a large language model, and a mechanical knowledge graph is automatically constructed based on the extracted entities and relationships based on a graph construction algorithm, thereby improving the efficiency and accuracy of entity and relationship extraction, thereby achieving efficient and stable construction of the mechanical knowledge graph.

[0069] Embodiment 2

[0070] Figure 3 A flowchart of a method for constructing a mechanical knowledge graph is provided in a second embodiment of the present invention. This embodiment is based on the above embodiment. After adopting a graph construction algorithm to connect each entity relationship subgraph to obtain a mechanical knowledge graph, it also includes: obtaining a mechanical maintenance request input by a user, and extracting a target entity contained in the mechanical maintenance request; querying the mechanical knowledge graph based on the target entity to obtain an associated entity corresponding to the target entity, and generating a maintenance suggestion based on the associated entity.

[0071] Step S201 , extracting entities and relationships from each mechanical knowledge document based on a large language model.

[0072] Optionally, entities and relationships are extracted from each mechanical knowledge document based on a large language model, including: extracting entities from each mechanical knowledge document based on a preset graph architecture based on the large language model, and deleting the extracted identical entities using an entity disambiguation method, wherein the entities include product line names, system names, component names, parameters, and fault types; extracting relationships corresponding to the entities from the mechanical knowledge documents, wherein the relationships include assembly relationships between components, faults occurring in components, and relationships between components and parameters.

[0073] Step S202: construct an entity relationship subgraph according to the entities and relationships extracted from each mechanical knowledge document.

[0074] Optionally, an entity relationship subgraph is constructed based on the entities and relationships extracted from each mechanical knowledge document, including: obtaining triples determined by the entities and relationships extracted from each mechanical knowledge document, wherein the triples include a relationship and a pair of entities; evaluating the relationship confidence in the triples, and filtering the triples according to the relationship confidence; and constructing an entity relationship subgraph corresponding to each mechanical knowledge document based on the filtered triples.

[0075] Optionally, evaluating the confidence of the relationship in the triples includes: calculating the frequency of occurrence of each relationship in the triples extracted from each mechanical knowledge document; and determining the confidence of the relationship according to the frequency of occurrence.

[0076] Step S203: Use a graph construction algorithm to connect the entity relationship subgraphs to obtain a mechanical knowledge graph.

[0077] Optionally, a graph construction algorithm is used to connect each entity relationship subgraph to obtain a mechanical knowledge graph, including: obtaining an expert knowledge tree, wherein the expert knowledge tree includes standard entities and relationships in the mechanical field; and using a graph construction algorithm to connect each entity relationship subgraph based on the expert knowledge tree to obtain a mechanical knowledge graph.

[0078] Optionally, a graph construction algorithm is used to connect each entity relationship subgraph based on the expert knowledge graph to obtain a mechanical knowledge graph, including: determining common entities in the expert knowledge tree and each entity relationship subgraph, wherein the common entities include the same entity or similar entities; using the common entities as alignment nodes, and connecting the expert knowledge tree and each entity relationship subgraph based on the alignment nodes to obtain a mechanical knowledge graph.

[0079] Optionally, determining common entities in the expert knowledge tree and each entity relationship subgraph includes: calculating the similarity between entities in the expert knowledge tree and each entity relationship subgraph; treating two entities whose similarity is equal to a first value as the same entity, and treating two entities whose similarity is greater than a second value and less than the first value as similar entities.

[0080] Step S204, obtaining the mechanical maintenance request input by the user, and extracting the target entity contained in the mechanical maintenance request.

[0081] Specifically, in this implementation, the mechanical knowledge graph constructed above can be applied to the intelligent question-and-answer application scenario of mechanical maintenance. For example, when the engine of a mechanical equipment fails, the maintenance personnel may not be particularly familiar with the structure of the mechanical equipment. At this time, a mechanical maintenance request "The engine fails, requesting maintenance suggestions" can be input to the equipment. When the mechanical maintenance request is received, the target entity "engine failure" will be extracted from it. In this implementation, the target entity will be detected to determine whether it is located in the mechanical knowledge graph. When it is determined that it is located in the mechanical knowledge graph, subsequent maintenance suggestions will be generated based on the target entity. However, when it is not located in the mechanical knowledge graph, a query failure prompt message will be generated and fed back to the user and the system, so that the system can update the mechanical knowledge graph based on the target entity.

[0082] It should be noted that in this embodiment, the number of target entities extracted from the mechanical maintenance request can be multiple, and the specific number of the extracted target entities is not limited in this embodiment. In addition, when extracting the target entities, the above-mentioned knowledge graph extraction tool can be used for extraction. Of course, deep learning can also be used for extraction. This embodiment does not limit the method of extracting the target entities.

[0083] Step S205: query the mechanical knowledge graph based on the target entity to obtain the associated entity corresponding to the target entity, and generate maintenance suggestions based on the associated entity.

[0084] Specifically, in this implementation, when the target entity is obtained and it is determined that the target entity is located in the created mechanical knowledge graph, the mechanical knowledge graph will be traversed to find the associated entity corresponding to the target entity, wherein the associated entity refers to the entity connected to the target entity in the mechanical knowledge graph, and the number of associated entities determined can be multiple, and the number of associated entities obtained is not limited in this implementation.

[0085] For example, the associated entities corresponding to the target entity "engine failure" include "connecting rod" and "crankshaft", and after obtaining the associated entities, maintenance suggestions can be generated based on the associated entities, such as "please check and repair the connecting rod and crankshaft". Therefore, even when the maintenance personnel are not clear about the internal structure of the equipment, they can quickly obtain the specific component structures associated with the current fault, thereby efficiently locating the fault and prompting the maintenance personnel so that they can perform mechanical maintenance based on the located fault point.

[0086] It should be noted that, in this implementation, the mechanical knowledge graph can also be applied to engineering machinery application scenarios such as intelligent reporting and fault analysis, and the specific application scenarios of the mechanical knowledge graph are not limited in this implementation.

[0087] In the implementation mode of the present application, for the engineering machinery knowledge management scenario, entities and relationships are extracted from diversified data in the machinery field based on a large language model, and a mechanical knowledge graph is automatically constructed based on the extracted entities and relationships based on a graph construction algorithm, thereby improving the efficiency and accuracy of entity and relationship extraction, thereby achieving efficient and stable construction of the mechanical knowledge graph.

[0088] Embodiment 3

[0089] Figure 4 This is a schematic diagram of the structure of a mechanical knowledge graph construction device provided in the third embodiment of the present invention. Figure 4 As shown, the device includes: an entity and relationship extraction module 310, an entity relationship subgraph construction module 320 and a mechanical knowledge graph acquisition module 330.

[0090] The entity and relationship extraction module 310 is used to extract entities and relationships from each mechanical knowledge document based on the large language model, wherein the mechanical knowledge document includes a maintenance work order, an overhaul manual, and a product manual;

[0091] An entity relationship subgraph construction module 320, for constructing an entity relationship subgraph according to entities and relationships extracted from each mechanical knowledge document;

[0092] The mechanical knowledge graph acquisition module 330 is used to connect the entity relationship subgraphs using a graph construction algorithm to obtain a mechanical knowledge graph.

[0093] Optionally, an entity and relationship extraction module is used to extract entities from each mechanical knowledge document based on a large language model according to a preset graph architecture, and delete the extracted identical entities using entity disambiguation, wherein the entities include product line names, system names, component names, parameters, and fault types;

[0094] Relationships corresponding to entities are extracted from mechanical knowledge documents, wherein the relationships include assembly relationships between components, failures occurring in components, and relationships between components and parameters.

[0095] Optionally, an entity-relationship subgraph construction module is used to obtain triples determined by entities and relationships extracted from each mechanical knowledge document, wherein the triples include a relationship and a pair of entities;

[0096] Evaluate the relationship confidence in the triples and filter the triples based on the relationship confidence;

[0097] The entity relationship subgraph corresponding to each mechanical knowledge document is constructed based on the filtered triples.

[0098] Optionally, the entity-relationship subgraph construction module is also used to calculate the frequency of occurrence of each relationship in the triples extracted from each mechanical knowledge document;

[0099] The confidence level of a relationship is determined based on the frequency of occurrence, where the frequency is proportional to the confidence level.

[0100] Optionally, a mechanical knowledge graph acquisition module is used to acquire an expert knowledge tree, wherein the expert knowledge tree includes standard entities and relationships in the mechanical field;

[0101] A graph construction algorithm is used to connect the entity relationship subgraphs based on the expert knowledge tree to obtain the mechanical knowledge graph.

[0102] Optionally, a mechanical knowledge graph acquisition module is used to determine common entities in the expert knowledge tree and each entity relationship subgraph, wherein the common entities include the same entity or similar entities;

[0103] The common entities are used as alignment nodes, and based on the alignment nodes, the expert knowledge tree and each entity relationship subgraph are connected to obtain the mechanical knowledge graph.

[0104] Optionally, a mechanical knowledge graph acquisition module is used to calculate the similarity between the expert knowledge tree and each entity in the entity relationship subgraph;

[0105] Two entities whose similarity is equal to the first value are regarded as the same entity, and two entities whose similarity is greater than the second value and less than the first value are regarded as similar entities.

[0106] Optionally, the device further comprises a maintenance suggestion generating module, which is used to obtain a mechanical maintenance request input by a user and extract a target entity contained in the mechanical maintenance request;

[0107] Based on the target entity, the mechanical knowledge graph is queried to obtain the associated entities corresponding to the target entity, and maintenance suggestions are generated based on the associated entities.

[0108] The mechanical knowledge graph construction device provided in the embodiment of the present invention can execute the mechanical knowledge graph construction method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0109] Embodiment 4

[0110] Figure 5A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0111] like Figure 5 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0112] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0113] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a mechanical knowledge graph construction method.

[0114] In some embodiments, the mechanical knowledge graph construction method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the mechanical knowledge graph construction method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured as a mechanical knowledge graph construction method in any other appropriate manner (e.g., by means of firmware).

[0115] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0116] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0117] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0118] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0119] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0120] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0121] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0122] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for constructing a mechanical knowledge graph, characterized in that: include: Extracting entities and relationships from each mechanical knowledge document based on a large language model, wherein the mechanical knowledge document includes a maintenance work order, an overhaul manual, and a product manual; constructing an entity-relationship subgraph according to the entities and relationships extracted from each of the mechanical knowledge documents; A graph construction algorithm is used to connect the entity relationship subgraphs to obtain a mechanical knowledge graph.

2. The method according to claim 1, characterized in that: The entity and relationship extraction of each mechanical knowledge document based on the large language model includes: Extracting entities from each of the mechanical knowledge documents based on a large language model according to a preset graph architecture, and deleting the extracted identical entities using an entity disambiguation method, wherein the entities include product line names, system names, component names, parameters, and fault types; Relationships corresponding to the entities are extracted from the mechanical knowledge document, wherein the relationships include assembly relationships between components, failures occurring in components, and relationships between components and parameters.

3. The method according to claim 1, characterized in that The step of constructing an entity-relationship subgraph based on the entities and relationships extracted from each of the mechanical knowledge documents comprises: Acquire a triple determined by entities and relations extracted from each mechanical knowledge document, wherein the triple includes a relation and a pair of entities; evaluating the relation confidences in the triples, and screening the triples according to the relation confidences; The entity relationship subgraph corresponding to each of the mechanical knowledge documents is constructed according to the filtered triples.

4. The method according to claim 3, characterized in that The evaluating the relationship confidence in the triplet comprises: Calculating the frequency of occurrence of each relationship in the triples extracted from each mechanical knowledge document; The confidence level of the relationship is determined based on the occurrence frequency.

5. The method according to claim 1, characterized in that The method of using a graph construction algorithm to connect the entity relationship subgraphs to obtain a mechanical knowledge graph includes: Acquire an expert knowledge tree, wherein the expert knowledge tree includes standard entities and relationships in the mechanical field; The graph construction algorithm is used to connect each of the entity relationship subgraphs based on the expert knowledge tree to obtain a mechanical knowledge graph.

6. The method according to claim 5, characterized in that The method of using the graph construction algorithm to connect the entity relationship subgraphs based on the expert knowledge graph to obtain a mechanical knowledge graph includes: Determine common entities in the expert knowledge tree and each of the entity relationship subgraphs, wherein the common entities include the same entity or similar entities; The common entity is used as an alignment node, and based on the alignment node, the expert knowledge tree and each entity relationship subgraph are connected to obtain the mechanical knowledge graph.

7. The method according to claim 6, characterized in that The determining of common entities in the expert knowledge tree and each of the entity relationship subgraphs comprises: Calculating the similarity between the expert knowledge tree and each entity in the entity relationship subgraph; The two entities whose similarity is equal to the first value are regarded as the same entity, and the two entities whose similarity is greater than the second value and less than the first value are regarded as the similar entities.

8. The method according to any one of claims 1 to 7, characterized in that: After the graph construction algorithm is used to connect the entity relationship subgraphs to obtain the mechanical knowledge graph, the method further includes: Obtaining a mechanical maintenance request input by a user, and extracting a target entity contained in the mechanical maintenance request; The mechanical knowledge graph is queried based on the target entity to obtain associated entities corresponding to the target entity, and maintenance suggestions are generated based on the associated entities.

9. A mechanical knowledge graph construction device, characterized in that: include: An entity and relationship extraction module, used for extracting entities and relationships from each mechanical knowledge document based on a large language model, wherein the mechanical knowledge document includes a maintenance work order, an overhaul manual, and a product manual; An entity-relationship subgraph construction module, used for constructing an entity-relationship subgraph according to entities and relationships extracted from each of the mechanical knowledge documents; The mechanical knowledge graph acquisition module is used to connect the entity relationship subgraphs by adopting a graph construction algorithm to obtain a mechanical knowledge graph.

10. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the method according to any one of claims 1 to 8.

11. 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 processor to implement the method according to any one of claims 1 to 8 when executed.

Images