Semi-automatic knowledge graph construction method and device and computer equipment
Patent Information
- Application Number
- CN202280095297.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2022-05-30
- Publication Date
- 2024-12-06
AI Technical Summary
In the process of knowledge graph construction, existing technology cannot ensure the accuracy of knowledge extraction and wastes a lot of human labor.
Through the historical fault case library based on industrial scenarios, phenomenon nodes, cause nodes and solution nodes are generated, and candidate nodes and relationships are displayed on the computer graphical user interface. Combined with algorithm recommendation and user confirmation, one by one is confirmed whether to be included in the map and retained, reducing Human intervention.
While ensuring the accuracy of the knowledge graph, it significantly reduces the amount of human labor in the construction process of the knowledge graph and improves efficiency.
Smart Images

Figure CN119096251A_ABST
Abstract
Description
Method, device, and computer equipment for semi-automatic knowledge graph construction Technical Field
[0001] The present application relates to natural language processing, and specifically, to a method, apparatus, computer device, and storage medium for semi-automatic knowledge graph construction. Background Art
[0002] A knowledge graph is a structured semantic knowledge base used to describe concepts and their relationships in the physical world in symbolic form. Its basic building blocks are "entity-relationship-entity" triples, along with entities and their associated attribute-value pairs. Entities are interconnected through relationships, forming a network-like knowledge structure. However, in many scenarios, knowledge extraction is inaccurate. Furthermore, significant labor is wasted in the process of generating knowledge graphs.
[0003] Summary of the Invention
[0004] The present application provides a method, comprising: generating a phenomenon node, a cause node, and a solution node based on a historical fault case library of industrial scenarios, wherein the phenomenon node indicates the state of the equipment when the fault occurs, the cause node indicates the event that caused the fault, and the solution node indicates how to resolve the event that caused the fault; displaying multiple candidate nodes on a computer graphical user interface for at least one of the phenomenon node, the cause node, and the solution node; and confirming one by one whether the multiple candidate nodes are included in the graph.
[0005] Using the above method, the accuracy of knowledge graph generation can be guaranteed while reducing the human labor in the process of generating the knowledge graph.
[0006] In one embodiment, the above method further includes: displaying multiple candidate relationships between the multiple candidate nodes and at least one of the phenomenon node, the cause node and the solution node on a computer graphical user interface; and confirming one by one whether the multiple candidate relationships are retained.
[0007] In one embodiment, confirming one by one whether the plurality of candidate nodes are included in the graph further includes confirming one by one whether the plurality of candidate nodes are included in the graph based on algorithm recommendation and user confirmation.
[0008] In one embodiment, selecting one of the multiple candidate relationships as the candidate relationship further includes: confirming whether the multiple candidate relationships are to be retained one by one based on algorithm recommendation and user confirmation.
[0009] In one embodiment, the algorithm recommendation may be based on the repetition of the contents of the multiple candidate nodes, judging the semantics of the multiple candidate nodes, and thereby merging the multiple candidate nodes.
[0010] In one embodiment, by determining that the content indicated by the solution node can eliminate the event causing the fault, one of the multiple candidate relationships is selected as the candidate relationship.
[0011] In one embodiment, at least one of the phenomenon node, the cause node, and the solution node is standardized, wherein the standardization includes creating or deleting candidate relationships between the multiple candidate nodes, and creating or deleting the multiple candidate nodes.
[0012] In one embodiment, the multiple candidate nodes of the knowledge graph are updated based on the user's second query input.
[0013] The present application also provides a device for semi-automatic knowledge graph construction, including: a generation module for generating phenomenon nodes, cause nodes and solution nodes based on a historical fault case library of industrial scenarios, wherein the phenomenon node indicates the status of the equipment when the fault occurs, the cause node indicates the event that causes the fault, and the solution node indicates how to solve the event that causes the fault; a display module for displaying multiple candidate nodes on a computer graphical user interface for at least one of the phenomenon node, the cause node and the solution node; and a selection module for confirming one by one whether the multiple candidate nodes are included in the graph.
[0014] The present application also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and wherein the processor implements the above method when executing the computer program.
[0015] The present application also provides a computer-readable storage medium having a computer program stored thereon, and the computer program implements the above method when executed by a processor. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] The following drawings are only intended to illustrate and explain the present application and do not limit the scope of the present application.
[0017] FIG1 is a flowchart of a method for semi-automatic knowledge graph construction according to an embodiment of the present application.
[0018] FIG2 is a schematic structural diagram of an apparatus for semi-automatic knowledge graph construction according to an embodiment of the present application.
[0019] FIG3 is a schematic structural diagram of a device for semi-automatic knowledge graph construction according to an embodiment of the present application.
[0020] The accompanying drawings are numerals as follows:
[0021] Steps S101-S103
[0022] 200: Device for semi-automatic knowledge graph construction
[0023] 201: Generate module
[0024] 202: Display module
[0025] 203: Select module
[0026] 300: Computer equipment
[0027] 302: Processor
[0028] 304: Memory DETAILED DESCRIPTION
[0029] In order to have a clearer understanding of the technical features, purposes and effects of this application, the specific implementation methods of this application are now described with reference to the accompanying drawings.
[0030] FIG1 is a flowchart of a method for semi-automatic knowledge graph construction according to an embodiment of the present application.
[0031] To provide solutions to problems in industrial scenarios, a knowledge graph can be used to describe the relationship between problems and solutions in industrial scenarios. The knowledge graph can be constructed based on a historical fault case library.
[0032] A knowledge graph consists of multiple nodes and their relationships. A node represents an atomic piece of knowledge about a fault's phenomenon, cause, or solution. A node can be a short sentence or a longer clause in a document in a fault history case library. Nodes can be categorized into the following types: phenomenon nodes, cause nodes, and solution nodes (hereinafter, phenomenon and phenomenon nodes, cause and cause nodes, and solution and solution nodes will be used interchangeably).
[0033] A phenomenon may refer to the state of the equipment at the time of the fault. For example, a coal feeder may be out of coal (hereinafter referred to as P1). A coal feeder is a mechanical device that accurately adjusts the amount of coal fed to the pulverizer according to load requirements. A pulverizer is a machine that breaks up coal lumps and grinds them into pulverized coal. In one embodiment, a phenomenon may be an abnormal noise emitted by the coal feeder belt (hereinafter referred to as P2). In another embodiment, a phenomenon may also be the pulverizer operating without a load (hereinafter referred to as P3).
[0034] The cause may refer to the condition that triggers the fault. For example, the coal feeder motor trips (hereinafter referred to as R1), meaning the power to the coal feeder's power unit is unintentionally disconnected. In another embodiment, the cause may be a coal feeder belt jam (hereinafter referred to as R2).
[0035] The solution can refer to how to solve the problem described in the phenomenon or the problem described in the cause. For example, the solution may be restarting the motor (hereinafter referred to as S1). In another embodiment, the solution may be replacing the motor (hereinafter referred to as S2). In another embodiment, the solution may be removing foreign objects near the coal feeder belt (hereinafter referred to as S3).
[0036] The following describes how to generate a knowledge graph. Generally speaking, knowledge graph generation involves first creating a small knowledge graph and then integrating it into an existing knowledge graph. The process of generating a small knowledge graph includes generating candidate nodes, confirming candidate nodes, and confirming candidate relationships. Integrating a small knowledge graph into an existing knowledge graph involves merging nodes and then associating them.
[0037] In one embodiment, the present application provides a method for semi-automatic knowledge graph construction, comprising: generating phenomenon nodes, cause nodes, and solution nodes based on a historical fault case library of industrial scenarios, wherein the phenomenon node indicates the state of the equipment when the fault occurs, the cause node indicates the event that causes the fault, and the solution node indicates how to solve the event that causes the fault; for at least one of the phenomenon node, the cause node, and the solution node, displaying multiple candidate nodes on a computer graphical user interface; and selecting one from the multiple candidate nodes as a candidate node.
[0038] Using the above method, the accuracy of knowledge graph generation can be guaranteed while reducing the human labor in the process of generating the knowledge graph.
[0039] The above process is described in detail below.
[0040] In step S101, based on a historical fault case library of industrial scenarios, phenomenon nodes, cause nodes and solution nodes are generated, wherein the phenomenon node indicates the state of the equipment when the fault occurs, the cause node indicates the event that causes the fault, and the solution node indicates how to solve the event that causes the fault.
[0041] The user inputs a "query input" into the retrieval-based diagnostic system. For example, the query input is "coal feeder failure." The retrieval system retrieves the most relevant historical case documents. Then, three target fields (phenomenon, cause, solution) are extracted from the historical case documents. Then, the segmentation method is used to decompose them into sentence units (SU). Then, the target field is labeled and multiple sentence units are aggregated. The specific method is not described in detail here, and can be found in PCT patent application PCT / CN2022 / 072778.
[0042] In step S102 , a plurality of candidate nodes are displayed on a computer graphical user interface for at least one of the phenomenon node, the cause node, and the solution node.
[0043] Candidate nodes can be generated from a historical fault case library. A historical fault case library can record the process of a fault. For example, a historical fault case library might contain the following text: "The coal feeder ran for two hours before it ran out of coal. Furthermore, after running for one hour, the feeder belt made an unusual noise." From this text, a candidate node can be generated that describes "the coal feeder ran for two hours."
[0044] Node confirmation refers to the confirmation of candidate nodes. For example, the user may not consider the above-mentioned node "Coal feeder runs for 2 hours" as a candidate node. In one embodiment, the above-mentioned node confirmation may be participated by the user. For example, the above-mentioned method may be implemented by software. The software has a user interface, which prompts the user to confirm: whether the candidate node can represent the atomic knowledge point of common faults; if it can, then whether the semantics of the candidate node is relevant to the query input. When the candidate node can represent the atomic knowledge point of common faults, the candidate node can be used as a candidate node, and the label is added to the existing knowledge graph. And when the semantics of the candidate node is relevant to the query input, the connection relationship between the candidate node and the existing knowledge graph will be confirmed. In one embodiment, when the user denies the relationship between the semantics of the candidate node and the query input, the connection relationship between the candidate node and the existing knowledge graph will be deleted. In one embodiment, the newly established node is not connected to the existing knowledge graph.
[0045] The following describes node merging. In one embodiment, the algorithm recommendation may be based on the repetition of the content of the multiple candidate nodes, and the semantics of the multiple candidate nodes may be judged, thereby merging the multiple candidate nodes. For example, candidate node 1 may be a coal feeder outage, and candidate node 2 may be a sudden decrease in the amount of coal in the coal feeder. Because candidate node 1 and candidate node 2 semantically represent the same content, they can be merged. There are many specific ways to determine whether the semantics of the candidate nodes are the same. For example, the semantics of the candidate nodes can be judged based on the repetition. In one embodiment, if the titles of the candidate nodes are similar enough, or most of the sentence units associated with the candidate nodes overlap, it can be determined that the semantics of the candidate nodes are the same.
[0046] Step S103: confirm one by one whether the multiple candidate nodes are included in the graph.
[0047] In one embodiment, multiple candidate relationships between the multiple candidate nodes and at least one of the phenomenon node, the cause node, and the solution node are displayed on a computer graphical user interface; and one of the multiple candidate relationships is selected as the candidate relationship. For example, the computer graphical user interface displays a connection between R1 and P1, and also displays a connection between R2 and P1, and the user can determine which of the above connections is to be selected as the relationship between P and R.
[0048] Using the above method, the accuracy of knowledge graph generation can be guaranteed while reducing the human labor in the process of generating the knowledge graph.
[0049] In another embodiment, it is possible to confirm one by one whether multiple candidate nodes are included in the graph based on algorithm recommendation and user confirmation. In another embodiment, it is possible to confirm one by one whether the multiple candidate relationships are retained based on algorithm recommendation and user confirmation. In the prior art, knowledge graphs are generated based on algorithm recommendation, or knowledge graphs are generated manually. The present application proposes a semi-automatic method, which uses a combination of algorithm and manual methods to generate knowledge graphs. Using the above method, the accuracy of the knowledge graph can be guaranteed while reducing the amount of human labor in the process of generating the knowledge graph.
[0050] In another embodiment, one of the multiple candidate relationships can be selected as the candidate relationship by determining whether the content indicated by the solution node can eliminate the event that caused the fault. For example, the user can determine whether S1 can eliminate the event in R1. In other words, the user can determine whether restarting the motor can resolve the coal feeder motor trip. If the coal feeder motor trip can be resolved, the connection relationship between S1 and R1 is confirmed. This semi-automatic approach, i.e., joint confirmation by the user and the algorithm, increases the accuracy of the knowledge graph and reduces the workload.
[0051] In another embodiment, at least one of the phenomenon node, the cause node, and the solution node is standardized, wherein the standardization includes creating or deleting candidate relationships between the multiple candidate nodes, and creating or deleting the multiple candidate nodes. For example, if the algorithm does not determine that P3 and P2 are duplicate candidate nodes, the user can delete phenomenon P3. In this way, the knowledge graph can be made more accurate.
[0052] In another embodiment, the plurality of candidate nodes of the knowledge graph are updated based on the user's second query input. For example, the user may input "coal feeder failure" and a candidate node "coal feeder motor trip" may be generated.
[0053] It should be understood that, although the various steps in the flowchart of FIG1 are shown in sequence as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in FIG2 may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.
[0054] Figure 2 provides a device 200 for semi-automatic knowledge graph construction, including: a generation module 201, which is used to generate phenomenon nodes, cause nodes and solution nodes based on a historical fault case library of industrial scenarios, wherein the phenomenon node indicates the status of the equipment when the fault occurs, the cause node indicates the event that causes the fault, and the solution node indicates how to solve the event that causes the fault; a display module 202, which is used to display multiple candidate nodes on a computer graphical user interface for at least one of the phenomenon node, the cause node and the solution node; and a selection module 203, which is used to confirm one by one whether the multiple candidate nodes are included in the graph.
[0055] In one embodiment, the device also includes a confirmation module configured to display multiple candidate relationships between the multiple candidate nodes and at least one of the phenomenon node, the cause node and the solution node on a computer graphical user interface; and confirm whether the multiple candidate relationships are retained one by one.
[0056] In one embodiment, the apparatus further comprises a confirmation module configured to confirm whether the candidate nodes are included in the graph one by one based on algorithm recommendation and user confirmation.
[0057] In one embodiment, the apparatus further comprises the confirmation module, further configured to confirm one by one whether the plurality of candidate relationships are retained.
[0058] In one embodiment, the apparatus further comprises a merging module configured to make a judgment on the semantics of the multiple candidate nodes based on the repetition of the contents of the multiple candidate nodes, thereby merging the multiple candidate nodes.
[0059] In one embodiment, the apparatus further includes: a selection module configured to select a relationship as a candidate node from the multiple candidate relationships by determining that the content indicated by the solution node can eliminate the event causing the fault.
[0060] In one embodiment, the device also includes: a standardization module configured to standardize at least one of the phenomenon node, the cause node and the solution node, wherein the standardization includes creating or deleting the candidate relationship between the multiple candidate nodes, and creating or deleting the multiple candidate nodes.
[0061] In one embodiment, the apparatus further includes: an updating module configured to update the multiple candidate nodes of the knowledge graph based on a second query input by the user.
[0062] For the specific definition of the device for semi-automatic knowledge graph construction, please refer to the definition of the method for semi-automatic knowledge graph construction above, which will not be repeated here. Each module in the above device can be implemented in whole or in part by software, hardware and a combination thereof. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.
[0063] Figure 3 provides a computer device for semi-automatic knowledge graph construction. According to one embodiment, the computer device 300 may include a processor 302 that executes a computer program stored in a memory 304. When the computer program is executed by the processor, a method for semi-automatic knowledge graph construction is implemented.
[0064] Those skilled in the art will understand that the structure shown in Figure 3 is merely a block diagram of a portion of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.
[0065] The present application also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the above steps when executed by a processor.
[0066] Those skilled in the art will appreciate that all or part of the processes in the methods for implementing the above embodiments can be accomplished by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).
[0067] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0068] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.
Claims
1. A method for semi-automatic knowledge graph construction, comprising: Based on the historical fault case library of industrial scenarios, generate phenomenon nodes, cause nodes and solution nodes, where the phenomenon node indicates the status of the equipment when the fault occurs, the cause node indicates the event that caused the fault, and the solution node indicates how to solve the event that caused the fault; For at least one of the phenomenon node, the cause node, and the solution node, displaying a plurality of candidate nodes on a computer graphical user interface; Confirm whether the multiple candidate nodes are included in the graph one by one.
2. The method according to claim 1, further comprising: displaying, on the computer graphical user interface, a plurality of candidate relationships between the plurality of candidate nodes and at least one of the phenomenon node, the cause node, and the solution node; It is confirmed one by one whether the plurality of candidate relationships are retained.
3. The method according to claim 1, wherein The step of confirming one by one whether the plurality of candidate nodes are included in the graph further includes confirming one by one whether the plurality of candidate nodes are included in the graph based on algorithm recommendation and user confirmation.
4. The method according to claim 2, wherein: Based on the algorithm recommendation and user confirmation, it is confirmed one by one whether the multiple candidate relationships are to be retained.
5. The method according to claim 3, wherein: The algorithm recommendation is based on the repetition of the contents of the multiple candidate nodes, and judges the semantics of the multiple candidate nodes, thereby merging the multiple candidate nodes.
6. The method according to claim 4, wherein: By determining that the content indicated by the solution node can eliminate the event causing the fault, one of the multiple candidate relationships is selected as the candidate relationship.
7. The method according to claim 5, wherein: At least one of the phenomenon node, the cause node, and the solution node is standardized, wherein the standardization includes creating or deleting the candidate relationship between the plurality of candidate nodes, and creating or deleting the plurality of candidate nodes.
8. The method according to claim 1, further comprising: Based on the user's second query input, the multiple candidate nodes of the knowledge graph are updated.
9. A device (200) for semi-automatically constructing a knowledge graph, comprising: A generation module (201) is used to generate a phenomenon node, a cause node and a solution node based on a historical fault case library of an industrial scenario, wherein the phenomenon node indicates the state of the device when the fault occurs, the cause node indicates the event that caused the fault, and the solution node indicates how to solve the event that caused the fault; a display module (202) for displaying a plurality of candidate nodes on a computer graphical user interface for at least one of the phenomenon node, the cause node, and the solution node; The selection module (203) is used to confirm whether the multiple candidate nodes are included in the graph one by one.
10. The apparatus according to claim 9, further comprising: a confirmation module configured to display, on the computer graphical user interface, a plurality of candidate relationships between the plurality of candidate nodes and at least one of the phenomenon node, the cause node, and the solution node; It is confirmed one by one whether the plurality of candidate relationships are retained.
11. The apparatus according to claim 9, further comprising: The confirmation module is configured to confirm whether the multiple candidate nodes are included in the graph one by one based on algorithm recommendation and user confirmation.
12. The device according to claim 10, characterized in that The confirmation module is further configured to confirm one by one whether the multiple candidate relationships are retained.
13. The apparatus according to claim 11, further comprising: The merging module is configured to make a judgment on the semantics of the multiple candidate nodes based on the repetition of the contents of the multiple candidate nodes, thereby merging the multiple candidate nodes.
14. The apparatus according to claim 12, further comprising: A selection module is configured to select one of the multiple candidate relationships as the candidate relationship by determining that the content indicated by the solution node can eliminate the event causing the fault.
15. The apparatus according to claim 13, further comprising: A standardization module is configured to standardize at least one of the phenomenon node, the cause node and the solution node, wherein the standardization includes creating or deleting the candidate relationship between the multiple candidate nodes, and creating or deleting the multiple candidate nodes.
16. The apparatus according to claim 9, further comprising: An updating module is configured to update the multiple candidate nodes of the knowledge graph based on the user's second query input.
17. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.
18. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.
