Knowledge graph updating methods, devices, electronic equipment and storage media

Abstract

This invention provides a knowledge graph updating method, apparatus, electronic device, and storage medium. The knowledge graph updating method includes: acquiring a first knowledge graph; constructing a second knowledge graph based on current fault data; updating a first weight in the first knowledge graph according to the second knowledge graph; and / or expanding the nodes and edges in the first knowledge graph. The above-described knowledge graph updating method of this invention improves the accuracy of the first knowledge graph by updating it, ensuring its dynamic updating and real-time performance.

Description

Technical Field

[0001] This invention relates to the field of knowledge graph technology, and in particular to a knowledge graph updating method, apparatus, electronic device, and storage medium. Background Technology

[0002] The automotive repair industry has always relied on the traditional apprenticeship model, which has resulted in problems such as a large amount of parts information, weak correlation between fault causes and faulty parts, and low level of structuring of vehicle fault record data. Therefore, the entire automotive repair industry urgently needs strong intelligent service support.

[0003] In existing technologies, a vehicle fault knowledge graph is constructed to record the relationship between fault phenomena and fault causes, so as to quickly and accurately analyze and locate the causes of vehicle faults based on the vehicle fault knowledge graph.

[0004] However, the inventors found that the construction of the vehicle fault knowledge graph requires users to predefine the knowledge graph, which relies too much on subjective human participation, and the content of the knowledge graph cannot be updated in real time, resulting in low accuracy of the vehicle fault knowledge graph. Summary of the Invention

[0005] In view of this, the present invention aims to provide a knowledge graph updating method, apparatus, storage medium, and vehicle to solve the technical problem of low accuracy of vehicle fault knowledge graphs in the prior art.

[0006] To achieve the above objectives, the technical solution of the present invention is implemented as follows:

[0007] The first aspect of this invention provides a knowledge graph updating method, comprising:

[0008] Obtain a first knowledge graph, wherein the first knowledge graph is constructed based on historical fault data, the first knowledge graph includes nodes, edges and a first weight for each edge, the nodes of the first knowledge graph include fault phenomenon nodes, fault cause nodes and fault component nodes, the edges of the first knowledge graph include a first edge connecting the fault phenomenon node and the fault cause node and a second edge connecting the fault cause node and the fault component node, the first weight of the first edge is used to represent the importance of the fault cause corresponding to the first edge, and the first weight of the second edge is used to represent the importance of the fault component corresponding to the second edge;

[0009] Construct a second knowledge graph based on the current fault data;

[0010] Based on the second knowledge graph, update the first weight in the first knowledge graph, and / or expand the nodes and edges in the first knowledge graph.

[0011] Optionally, the second knowledge graph includes nodes and edges, and the nodes of the second knowledge graph include fault phenomenon nodes, fault cause nodes, and fault component nodes;

[0012] The step of updating the first weights between nodes in the first knowledge graph based on the second knowledge graph, and / or expanding the nodes and edges in the first knowledge graph, includes:

[0013] Multiple first vector expressions are extracted from the nodes and edges of the first knowledge graph, wherein the first vector expression includes the corresponding fault phenomenon, fault cause and fault component;

[0014] Multiple second vector expressions are extracted from the nodes and edges of the second knowledge graph, wherein the second vector expressions include the corresponding fault phenomena, fault causes and faulty components;

[0015] Determine the degree of matching between the second vector expression and the first vector expression;

[0016] If the matching degree between the second vector expression and the first vector expression is greater than or equal to the preset matching value, then the first weight in the first knowledge graph is updated according to the second vector expression;

[0017] If the matching degree between the second vector expression and all the first vector expressions is less than the preset matching value, then the nodes and edges in the first knowledge graph are expanded according to the second vector expression, or the nodes and edges in the first knowledge graph are expanded and the first weight in the first knowledge graph is updated.

[0018] Optionally, the step of updating the first weights between nodes in the first knowledge graph and / or expanding the nodes and edges in the first knowledge graph according to the second knowledge graph further includes:

[0019] A first cause vector expression is extracted from the fault phenomenon nodes and fault cause nodes connected by edges in the first knowledge graph, wherein the first cause vector expression includes the corresponding fault phenomenon and fault cause.

[0020] The second cause vector expression is extracted from the fault phenomenon nodes and fault cause nodes connected by edges in the second knowledge graph, wherein the second cause vector expression includes the corresponding fault phenomenon and fault cause.

[0021] Determine the degree of matching between the second cause vector expression and the first cause vector expression;

[0022] If the matching degree between the second cause vector expression and the first cause vector expression is greater than or equal to the matching preset value, then the first weight in the first knowledge graph is updated according to the second cause vector expression;

[0023] If the matching degree between the second cause vector expression and all the first cause vector expressions is less than the matching preset value, then the nodes and edges in the first knowledge graph are expanded according to the second cause vector expression, or the nodes and edges in the first knowledge graph are expanded and the first weight in the first knowledge graph is updated.

[0024] Optionally, updating the first weights between nodes in the first knowledge graph based on the second knowledge graph, and / or expanding the nodes and edges in the first knowledge graph, further includes:

[0025] The first component vector expression is extracted from the fault cause nodes and fault component nodes connected by edges in the first knowledge graph, wherein the first component vector expression includes the corresponding fault cause and fault component.

[0026] The second component vector expression is extracted from the fault cause nodes and fault component nodes connected by edges in the second knowledge graph, wherein the second component vector expression includes the corresponding fault cause and fault component;

[0027] Determine the matching degree between the second component vector expression and the first component vector expression;

[0028] If the matching degree between the second component vector expression and the first component vector expression is greater than or equal to the matching preset value, then the first weight in the first knowledge graph is updated according to the second component vector expression;

[0029] If the matching degree between the second component vector expression and all the first component vector expressions is less than a preset matching value, then the nodes and edges in the first knowledge graph are expanded according to the second component vector expression, or the nodes and edges in the first knowledge graph are expanded and the first weight in the first knowledge graph is updated.

[0030] Optionally, the edges in both the first and second knowledge graphs have attributes, with the edge attributes in the first knowledge graph including a first number and the edge attributes in the second knowledge graph including a second number.

[0031] The step of updating the first weight in the first knowledge graph according to the second vector expression includes:

[0032] Update the edge attributes of the edge corresponding to the first vector expression based on the edge attributes of the edge corresponding to the second vector expression;

[0033] Update the first weight in the first knowledge graph based on the edge attributes of the edge corresponding to the updated first vector expression.

[0034] Optionally, before the step of obtaining the first knowledge graph, the method further includes:

[0035] Acquire historical fault data, wherein the historical fault data includes multiple fault information of the same vehicle model within a historical time period, and each fault information includes the corresponding fault phenomenon, fault cause and fault component;

[0036] Based on the historical fault data, determine the weight of the fault cause in its corresponding fault phenomenon;

[0037] Based on the historical fault data, the weight of the faulty component in its corresponding fault cause is determined;

[0038] A first knowledge graph is constructed based on the historical fault data, the weight of the fault cause in its corresponding fault phenomenon, and the weight of the faulty component in its corresponding fault cause.

[0039] Optionally, constructing a second knowledge graph based on the current fault data includes:

[0040] Acquire current fault data, wherein the current fault data includes fault cause analysis information and processing information for the same vehicle model within the current time period;

[0041] Extract the cause triplet between the fault phenomenon and the fault cause from the fault cause analysis information;

[0042] Extract the component triplet between the fault cause and the faulty component from the processing information;

[0043] A second knowledge graph is constructed based on the cause triples and the component triples.

[0044] A second aspect of this invention provides a knowledge graph updating apparatus, the apparatus comprising:

[0045] The first knowledge graph acquisition module is used to acquire a first knowledge graph, wherein the first knowledge graph is constructed based on historical fault data. The first knowledge graph includes nodes, edges, and a first weight for each edge. The nodes of the first knowledge graph include fault phenomenon nodes, fault cause nodes, and fault component nodes. The edges of the first knowledge graph include a first edge connecting the fault phenomenon node and the fault cause node and a second edge connecting the fault cause node and the fault component node. The first weight of the first edge is used to represent the importance of the fault cause corresponding to the first edge, and the first weight of the second edge is used to represent the importance of the fault component corresponding to the second edge.

[0046] The second knowledge graph construction module is used to construct a second knowledge graph based on the current fault data;

[0047] The update module is used to update the first weight in the first knowledge graph based on the second knowledge graph, and / or expand the nodes and edges in the first knowledge graph.

[0048] A third aspect of the present invention provides an electronic device, comprising:

[0049] processor;

[0050] Memory used to store the processor's executable instructions;

[0051] The processor is configured to execute the instructions to implement the knowledge graph update method as described in any of the first aspects.

[0052] A fourth aspect of the present invention provides a readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the knowledge graph update method described in any of the first aspects of the present invention.

[0053] Compared with existing technologies, the knowledge graph updating method, apparatus, electronic device, and storage medium provided by this invention have the following advantages:

[0054] In this embodiment of the invention, the first weight in the first knowledge graph is updated based on the second knowledge graph, and / or the nodes and edges in the first knowledge graph are expanded. That is, the first knowledge graph, after being constructed based on historical fault data, is not static but continuously updated and optimized. Updating the first knowledge graph improves its accuracy, avoiding the decrease in accuracy caused by untimely updates, thus solving the lag issue and ensuring the dynamic updating and real-time performance of the first knowledge graph. Furthermore, the first knowledge graph also includes a first weight for each edge. Setting the first weight of an edge reflects the strength of the association between the two nodes corresponding to that edge, making the fault diagnosis results derived from the first knowledge graph more accurate and facilitating rapid fault diagnosis by maintenance personnel. Additionally, the first knowledge graph in this embodiment can be automatically constructed based on historical fault data, without requiring users to predefine the knowledge graph, i.e., without relying on subjective human intervention. Attached Figure Description

[0055] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings:

[0056] Figure 1 A flowchart illustrating the steps of a knowledge graph updating method provided in this embodiment of the invention;

[0057] Figure 2 A schematic diagram of the first knowledge graph provided in the embodiments of the present invention. Figure 1 ;

[0058] Figure 3 A schematic diagram of the first knowledge graph provided in the embodiments of the present invention. Figure 2 ;

[0059] Figure 4 A schematic diagram of the second knowledge graph provided in an embodiment of the present invention;

[0060] Figure 5 This is a schematic diagram of a knowledge graph updating device provided in an embodiment of the present invention. Specific Implementation

[0061] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] In existing technologies, vehicle fault knowledge graphs constructed through user-defined methods include fault phenomenon nodes and fault cause nodes, and can represent the association between fault phenomenon nodes and fault cause nodes. However, the content of the knowledge graph cannot be updated in real time after its construction, resulting in low accuracy of the vehicle fault knowledge graph. For example, in a vehicle fault knowledge graph built based on the past year, the fault phenomenon node includes "the window cannot be closed," but its associated fault cause node is only "the wire is broken." However, according to recent repair processes, other causes of the window not closing include a dead battery and a damaged controller. Since the previously constructed vehicle fault knowledge graph has not been updated, searching for the cause of the window not closing based on this vehicle fault knowledge graph will only find a broken wire, which is inconsistent with the actual causes of the window not closing, such as a broken wire, a dead battery, and a damaged controller. Therefore, the vehicle fault knowledge graphs constructed in existing technologies have low accuracy and are subject to lag.

[0063] To address the aforementioned problems, this application proposes a knowledge graph updating method, apparatus, electronic device, and storage medium.

[0064] Firstly, referring to Figure 1 This document illustrates a flowchart of a knowledge graph updating method provided by an embodiment of the present invention. The knowledge graph updating method provided by this embodiment can be applied to terminals, which may be, but are not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The knowledge graph updating method includes:

[0065] Step 101: Obtain the first knowledge graph.

[0066] In this embodiment, a knowledge graph describes concepts, entities, and their relationships in the objective world in a structured form, expressing information in a way that more closely resembles human cognition of the world, and providing a better ability to organize, manage, and understand massive amounts of information. A knowledge graph is a structured semantic knowledge base, which can be understood as a semantic network graph composed of interconnected knowledge points. A knowledge graph can be displayed in the form of a graph, which is an abstract data structure used to represent the relationships between objects, described using nodes and edges. Nodes represent objects, i.e., entities; edges represent relationships or attributes between objects.

[0067] The first knowledge graph is built based on historical fault data. It only needs to be built once, and once completed, it can be accessed immediately. The first knowledge graph includes nodes, edges, and a weight for each edge. Each node corresponds to an entity; an edge can connect two entities, representing the association between them—that is, the relationship or attribute between the two entities. The edge weight represents the degree of association between the two entities; the higher the weight, the stronger the association. The first knowledge graph can be a vehicle fault knowledge graph, meaning the causes of vehicle faults can be obtained from it. When displaying the causes of vehicle faults to repair personnel through the first knowledge graph, results with higher relevance can be recommended to them.

[0068] Historical fault data includes multiple fault information for the same vehicle model within a historical time period. The vehicle model refers to its variant, such as the Haval H6 or ORA White Cat. Due to differences in internal components and construction, a first knowledge graph needs to be built for each vehicle model. The historical time period is a relatively long period of time that has passed, different from the current time. The duration of the historical time period can be several years, such as two or three years. For example, if the current time is September 1, 2022, then the historical time period could be from June 30, 2020 to June 30, 2022.

[0069] Both the first and second knowledge graphs contain nodes representing fault phenomena, fault causes, and faulty components. Fault phenomenon nodes and fault cause nodes are connected by edges in both graphs, as are fault cause nodes and faulty component nodes. Specifically, the first knowledge graph has edges connecting fault phenomenon nodes and fault cause nodes, and edges connecting fault cause nodes and faulty component nodes. The second knowledge graph has edges connecting fault phenomenon nodes and fault cause nodes, and edges connecting fault cause nodes and faulty component nodes.

[0070] Reference Figure 2 , Figure 2 A schematic diagram of the first knowledge graph is shown. Figure 1 Nodes have node types, and different nodes can have different node types. In this invention, the nodes in the first and second knowledge graphs are divided into three types: fault phenomena, fault causes, and faulty components. That is, the nodes in both the first and second knowledge graphs include fault phenomenon nodes, fault cause nodes, and faulty component nodes. There is an association between fault phenomenon nodes and fault cause nodes, and there is an association between fault cause nodes and faulty component nodes. For example... Figure 3The "window cannot be closed" node is a symptom node of a fault. Root cause nodes include a dead battery, a broken wire, a damaged controller, and worn-out rubber. Faulty components include the battery, wiring connections, internal wiring, the controller, and rubber. A symptom is the physical manifestation of a vehicle malfunction, such as a window that cannot be closed or an engine that cannot start. A cause is the reason that leads to the symptom; for example, a dead battery or a broken wire could cause a window to remain open. A faulty component is the part of the vehicle that is malfunctioning in relation to the cause of the malfunction; for example, if the battery is dead, the faulty component is the battery itself.

[0071] For the same fault symptom, there may be multiple corresponding causes. Different causes are more likely to cause the fault, and this likelihood indicates the relative importance of each cause. For example, for the fault symptom of a car window failing to close, a broken wire is more likely to cause the window to fail to close, while rubber aging is less likely to cause it. This indicates that the broken wire is more important than the aging rubber. Similarly, for the same fault symptom, there may be multiple corresponding faulty components. Different faulty components are more likely to cause the fault, and this likelihood indicates the relative importance of each component. For example, for the fault symptom of a broken wire, the wire connection is more likely to cause the break, while the internal structure of the wire is less likely to cause it.

[0072] In both the first and second knowledge graphs, the first weight of an edge represents the magnitude of the aforementioned probability. The first weight of the first edge in the first knowledge graph indicates the strength of the correlation between the fault cause and the fault phenomenon corresponding to the first edge, thus representing the importance of the fault cause corresponding to the first edge. Similarly, the first weight of the second edge in the first knowledge graph indicates the strength of the correlation between the faulty component and the fault cause corresponding to the second edge, thus representing the importance of the faulty component corresponding to the second edge. The second knowledge graph may also include a second weight for each edge. For example, the second weight of the third edge in the second knowledge graph indicates the strength of the correlation between the fault cause and the fault phenomenon corresponding to the third edge, thus representing the importance of the fault cause corresponding to the third edge. Likewise, the second weight of the fourth edge in the second knowledge graph indicates the strength of the correlation between the faulty component and the fault cause corresponding to the fourth edge, thus representing the importance of the faulty component corresponding to the fourth edge. Of course, each edge in the second knowledge graph may also be unweighted.

[0073] In this embodiment of the invention, the nodes in the first knowledge graph are subdivided into fault phenomenon nodes, fault cause nodes, and fault component nodes. Based on the first knowledge graph, the fault cause and fault component when a vehicle malfunctions can be quickly located, enabling targeted inspection and repair. This helps repair personnel quickly diagnose and efficiently resolve problems, greatly improving work efficiency and thus enhancing the efficiency of vehicle diagnostics. Furthermore, since edges have a first weight, inspection can be performed on fault causes and fault components with higher first weights, further improving the efficiency of vehicle diagnostics.

[0074] Step 102: Construct a second knowledge graph based on the current fault data.

[0075] In this embodiment, the current fault data includes fault data for the same vehicle model within the current time period, and multiple second knowledge graphs are constructed for each vehicle model. The current time period is a time period relatively close to the current time. The duration of the current time period can be several weeks or several days. For example, if the duration of the current time period can be one month, then when the current time is August 1, 2022, the current time period can be from July 1, 2022 to July 31, 2022.

[0076] There can be multiple second knowledge graphs. For example, a second knowledge graph can be constructed using the current fault data within the time period from July 1, 2022 to July 31, 2022. After updating the first knowledge graph constructed using the historical fault data within the time period from June 30, 2020 to June 30, 2022 with the second knowledge graph, when the current time is September 1, 2022, another second knowledge graph can be constructed using the current fault data within the time period from August 1, 2022 to August 31, 2022. The updated first knowledge graph can be updated with the second knowledge graph, so as to realize the continuous updating of the first knowledge graph. That is, steps 102 and 103 can be executed cyclically.

[0077] Step 103: Update the first weight in the first knowledge graph based on the second knowledge graph, and / or expand the nodes and edges in the first knowledge graph.

[0078] In this embodiment, when updating the first knowledge graph based on the second knowledge graph, the nodes and edges in the second knowledge graph may all already exist in the first knowledge graph. In this case, it is only necessary to update the first weight of the edges in the first knowledge graph. The nodes and edges in the second knowledge graph may also not exist in the first knowledge graph at all. In this case, it is necessary to expand the nodes and edges in the first knowledge graph. The edges and nodes in the second knowledge graph may also partially exist in the first knowledge graph. For example, an edge and two nodes adjacent to the edge may have one node that already exists in the first knowledge graph and the other node that does not exist in the first knowledge graph. In this case, it is necessary to expand the nodes and edges in the first knowledge graph and update the first weight in the first knowledge graph.

[0079] For example, refer to Figure 3 and Figure 4 , Figure 4 The second knowledge graph shown includes a node where the car window cannot be closed and a node associated with that node where the battery is dead. The nodes where the car window cannot be closed and the node where the battery is dead are connected by an edge. Figure 3 The first knowledge graph shown also includes a node indicating that the window cannot be closed and a node associated with that node indicating that the battery is dead. Therefore, the weight of the edge connecting the nodes indicating that the window cannot be closed and the dead battery node needs to be updated. The second knowledge graph includes a node indicating that the motor is damaged and a node associated with that node. The damaged motor node and the motor node are connected by an edge. The first knowledge graph does not include the damaged motor node or the motor node at all. Therefore, the damaged motor node and the motor node need to be added to the first knowledge graph, i.e., the nodes and edges in the first knowledge graph need to be expanded.

[0080] In this embodiment of the invention, the first weight in the first knowledge graph is updated based on the second knowledge graph, and / or the nodes and edges in the first knowledge graph are expanded. That is, the first knowledge graph, after being constructed based on historical fault data, is not static but continuously updated and optimized. Updating the first knowledge graph improves its accuracy, avoiding the decrease in accuracy caused by untimely updates, thus solving the lag issue and ensuring the dynamic updating and real-time performance of the first knowledge graph. Furthermore, the first knowledge graph also includes a first weight for each edge. Setting the first weight of an edge reflects the strength of the association between the two nodes corresponding to that edge, making the fault diagnosis results derived from the first knowledge graph more accurate and facilitating rapid fault diagnosis by maintenance personnel. Additionally, the first knowledge graph in this embodiment can be automatically constructed based on historical fault data, without requiring users to predefine the knowledge graph, i.e., without relying on subjective human intervention.

[0081] Specifically, step 103 includes:

[0082] Multiple first vector expressions are extracted from the nodes and edges of the first knowledge graph.

[0083] Multiple second vector expressions are extracted from the nodes and edges of the second knowledge graph.

[0084] In this embodiment, both the first vector expression and the second vector expression can be textual expressions. The first vector expression includes the corresponding fault phenomenon, fault cause, and faulty component, and the second vector expression includes the corresponding fault phenomenon, fault cause, and faulty component. In the first and second knowledge graphs, the fault phenomenon can be obtained from the fault phenomenon node, such as the fault phenomenon that the car window cannot be closed can be obtained from the node that the car window cannot be closed; the fault cause node can be obtained from the fault cause node, and the faulty component node can be obtained from the faulty component node. When a fault phenomenon node and a fault cause node are connected by an edge, the fault phenomenon obtained from the fault phenomenon node corresponding to that edge corresponds to the fault cause obtained from the fault cause node corresponding to that edge. Similarly, when a fault cause node and a faulty component node are connected by an edge, the fault cause obtained from the fault cause node corresponding to that edge corresponds to the faulty component obtained from the faulty component node corresponding to that edge.

[0085] Both the first and second knowledge graphs are represented using a tree-like structure, which includes a root node, child nodes connected to the root node, and leaf nodes connected to the child nodes. Of course, the first knowledge graph can also be represented using a conventional network structure, and the second knowledge graph can also be represented using a conventional network structure. In this case, the edges in both the first and second knowledge graphs can be directed edges.

[0086] Both the first and second vector expressions can be extracted in the order of root node, child node, and leaf node, or they can be extracted from nodes according to the direction of directed edges. Since both the first and second knowledge graphs contain multiple nodes and multiple edges, there will be multiple extracted first and second vector expressions. For example, for... Figure 3 The first knowledge graph shown is a tree structure, where the node "window cannot be closed" is the root node, the nodes "battery dead", "wire broken", "controller damaged" and "rubber aged" are child nodes, and the nodes "battery", "wire interface", "wire interior", "controller" and "rubber" are leaf nodes. The first vector expression can include three node information: root node, child node and leaf node, or it can only include the root node and child node, or only the child node and leaf node. In this case, the first vector expression can include (window cannot be closed, battery dead, battery), (window cannot be closed, wire broken, wire interface), or (window cannot be closed, battery dead) and (battery dead, battery).

[0087] Determine the degree of matching between the second vector expression and the first vector expression.

[0088] In this embodiment, the matching degree between the second vector expression and the first vector expression can be determined by text recognition and comparison. That is, the higher the overlap between the second vector expression and the first vector expression, the higher the matching degree. It should be noted that when determining the matching degree between the second vector expression and the first vector expression, word segmentation and matching must be performed sequentially.

[0089] For example, if the first vector expression is (the window cannot be closed, the battery is dead, the battery), and the second vector expression is (the window cannot be closed, the battery is dead, the battery), then the matching degree between the first and second vector expressions is 100%; if the first vector expression is (the window cannot be closed, the wire is broken), and the second vector expression is (the window cannot be closed, the motor is damaged), then the matching degree between the first and second vector expressions is 60%.

[0090] If the matching degree between the second vector expression and the first vector expression is greater than or equal to the preset matching value, then the first weight in the first knowledge graph is updated according to the second vector expression.

[0091] In this embodiment, the matching preset value can be set according to actual needs, such as 90%. When the matching degree between the second vector expression and any first vector expression is greater than or equal to the matching preset value, it means that the second vector expression and the first vector expression have a high degree of similarity. In this case, the content of the nodes they correspond to is basically the same, so there is no need to add nodes and edges, and only the first weight in the first knowledge graph needs to be updated.

[0092] If the matching degree between the second vector expression and all the first vector expressions is less than the preset matching value, then expand the nodes and edges in the first knowledge graph according to the second vector expression, or expand the nodes and edges in the first knowledge graph and update the first weight in the first knowledge graph.

[0093] In this embodiment, if the matching degree between the second vector expression and all the first vector expressions is less than a preset matching value, it indicates that there are no or only partially existing nodes and edges corresponding to the second vector expression in the first knowledge graph. In this case, it is necessary to expand the nodes and edges in the first knowledge graph. If there are no nodes and edges corresponding to the second vector expression in the first knowledge graph at all, it is only necessary to expand the nodes and edges in the first knowledge graph. If there are some nodes and edges corresponding to the second vector expression in the first knowledge graph, after expanding the nodes and edges in the first knowledge graph, it is also necessary to update the first weight of the edges in the first knowledge graph accordingly.

[0094] In this embodiment, by using the matching degree between the second vector expression and the first vector expression, it is possible to determine whether it is necessary to expand the nodes and edges in the first knowledge graph according to the second vector expression, or to update the first weight in the first knowledge graph according to the second vector expression. This method ensures the update efficiency of the knowledge graph, and does not require any node in the second knowledge graph to be directly added to the first knowledge graph, thus ensuring the simplicity of the first knowledge graph.

[0095] Step 103 may also include:

[0096] The first cause vector expression is extracted from the fault phenomenon nodes and fault cause nodes connected by edges in the first knowledge graph.

[0097] The second cause vector expression is extracted from the fault phenomenon nodes and fault cause nodes connected by edges in the second knowledge graph.

[0098] In this embodiment, the first vector expression can be divided into a first cause vector expression and a first component vector expression, and the second vector expression can be divided into a second cause vector expression and a second component vector expression. Both the first and second cause vector expressions include corresponding fault phenomena and fault causes, and both the first and second component vector expressions include corresponding fault causes and faulty components. For example, regarding... Figure 3 We can extract the first cause vector expression (window cannot be closed, battery is dead) and the first component vector expression (battery is dead, battery is dead) from the nodes connected by edges, namely, the window cannot be closed and the battery is dead. The extraction methods for the second cause vector expression and the second component vector expression are the same as those for the first cause vector expression, and will not be repeated here.

[0099] Determine the degree of matching between the second cause vector expression and the first cause vector expression.

[0100] In this embodiment, the method for determining the matching degree between the second cause vector expression and the first cause vector expression can refer to the method for determining the matching degree between the second vector expression and the first vector expression, and will not be repeated here.

[0101] If the matching degree between the second cause vector expression and the first cause vector expression is greater than or equal to the preset matching value, then the first weight in the first knowledge graph is updated according to the second cause vector expression.

[0102] In this embodiment, the matching preset value can be set according to actual needs, such as 90%. The first weight in the first knowledge graph that needs to be updated includes the first weight of all edges connecting the fault phenomenon node corresponding to the first cause vector expression for which the matching degree with the second cause vector expression is greater than or equal to the matching preset value. For example, if the first cause vector expression is (the window cannot be closed, the battery is dead), then the first weight in the first knowledge graph that needs to be updated includes the first weight of all edges connecting the node corresponding to the first cause vector expression where the window cannot be closed. The first weight in the first knowledge graph is updated according to the second cause vector expression, that is, according to the fault phenomenon node, the fault cause node, and the edges connecting the fault phenomenon node and the fault cause node corresponding to the second cause vector expression.

[0103] If the matching degree between the second cause vector expression and all the first cause vector expressions is less than the preset matching value, then the nodes and edges in the first knowledge graph are expanded according to the second cause vector expression, or the nodes and edges in the first knowledge graph are expanded and the first weight in the first knowledge graph is updated.

[0104] In this embodiment, the nodes and edges in the first knowledge graph are expanded according to the second cause vector expression. That is, the nodes and edges in the first knowledge graph are expanded according to the fault phenomenon node, the fault cause node, and the edge connecting the fault phenomenon node and the fault cause node corresponding to the second cause vector expression. For example, if the second cause vector expression is (wiper cannot be used, wiper controller is damaged), then the nodes and edges corresponding to (wiper cannot be used, wiper controller is damaged) are added to the first knowledge graph, that is, the nodes and edges in the first knowledge graph are expanded.

[0105] It should be noted that adding the aforementioned nodes and edges to the first knowledge graph requires adding each node individually. For nodes already existing in the first knowledge graph, no additional nodes are needed. For example, if the second cause vector expression is "(window cannot be closed, motor is damaged)," since the "window cannot be closed" node already exists in the first knowledge graph, this node does not need to be added. Only the "motor is damaged" node and the edge connecting it to both nodes need to be added. Furthermore, if the second cause vector expression is "(window cannot be closed, motor is damaged)," and the "window cannot be closed" node already exists in the first knowledge graph, after adding the "motor is damaged" node and the edge connecting it, the first weight of all edges connecting to the "window cannot be closed" node also needs to be updated.

[0106] In this embodiment, the cause vector expression is determined based on the fault phenomenon node and the fault cause node. The obtained cause vector expression can quickly locate the relevant edges, which facilitates the updating of the first weight of the edges.

[0107] Specifically, step 103 may also include:

[0108] Extract the first component vector expression from the fault cause nodes and fault component nodes connected by edges in the first knowledge graph; extract the second component vector expression from the fault cause nodes and fault component nodes connected by edges in the second knowledge graph; determine the matching degree between the second component vector expression and the first component vector expression; if the matching degree between the second component vector expression and the first component vector expression is greater than or equal to a preset matching value, update the first weight in the first knowledge graph according to the second component vector expression; if the matching degree between the second component vector expression and all the first component vector expressions is less than a preset matching value, expand the nodes and edges in the first knowledge graph according to the second component vector expression.

[0109] In this embodiment, the first weight in the first knowledge graph that needs to be updated based on the second component vector expression includes, for a first component vector expression whose matching degree with the second component vector expression is greater than or equal to a preset matching value, the first weight of all edges connecting the fault cause node corresponding to the first component vector expression. This embodiment is similar to the above embodiment of updating the first knowledge graph based on the first cause vector expression and the second cause vector expression, and can be referred to the above embodiment, so it will not be repeated here. In this embodiment, the component vector expression is determined based on the fault cause node and the fault component node. Based on the obtained component vector expression, the relevant edges can be quickly located, facilitating the updating of the first weight of the edges.

[0110] Specifically, edges in both the first and second knowledge graphs have attributes. Edges in the first knowledge graph have an "early count," while edges in the second knowledge graph have a "second count." When an edge in the first knowledge graph connects a fault phenomenon node to a fault cause node, the "early count" refers to the number of times that fault cause occurs among the causes leading to the fault phenomenon. When an edge in the first knowledge graph connects a fault cause to a faulty component, the "early count" refers to the number of times that faulty component fails among the components that fail for that fault cause.

[0111] For example, refer to Figure 3For the edge connecting the window that cannot be closed and the wire break node, the edge attribute is 10, which means that the wire break occurred 10 times among the reasons that the window cannot be closed; for the edge connecting the wire break node and the wire interface node, the edge attribute is 8, which means that the wire interface failed 8 times among the faulty components due to the wire break.

[0112] Update the first weight in the first knowledge graph according to the second vector expression, including:

[0113] Update the edge attributes of the edge corresponding to the first vector expression based on the edge attributes of the edge corresponding to the second vector expression.

[0114] In this embodiment, the edge attributes of the edges corresponding to the second vector expression can be determined based on the second knowledge graph. Based on the edge attributes of the edges corresponding to the second vector expression, the edge attributes of the edges corresponding to the first vector expression are updated, that is, the edge attributes of the edges corresponding to the second vector expression are added to the edge attributes of the edges corresponding to the first vector expression.

[0115] For example, refer to Figure 3 and Figure 4 When the second vector expression is (the car window cannot be closed, the battery is dead), the edge attribute of the edge corresponding to the second vector expression is 1. The first vector expression with a matching degree greater than or equal to the matching preset value is (the car window cannot be closed, the battery is dead), and the edge attribute of the edge corresponding to the first vector expression is 2. Then, the edge attribute 1 of the edge corresponding to the second vector expression is added to the edge attribute 2 of the edge corresponding to the first vector expression. The updated edge attribute of the edge corresponding to the first vector expression is 3.

[0116] Update the first weight in the first knowledge graph based on the edge attributes of the edge corresponding to the updated first vector expression.

[0117] In this embodiment, the first weight of an edge in the first knowledge graph is related to its edge attributes. For any first edge connecting a fault phenomenon node and a fault cause node in the first knowledge graph, the first weight is the ratio of the edge attribute of the first edge connecting the fault cause node and the fault phenomenon node to the sum of the edge attributes of all first edges connecting the fault phenomenon node. That is, based on the edge attributes corresponding to the updated first vector expression, the first weight of the relevant edges in the first knowledge graph can be updated. For example, Figure 3In the first knowledge graph, for the nodes "Window Cannot Be Closed" and "Battery Dead," the edge attribute of the first edge connecting the "Battery Dead" node and the "Window Cannot Be Closed" node is 2. The sum of the edge attributes of all the first edges connecting the "Window Cannot Be Closed" node is 2 + 10 + 7 + 1, which equals 20. The ratio of 2 to 20 is 0.1, therefore the first weight of the first edge connecting the "Window Cannot Be Closed" node and the "Battery Dead" node is 0.1. For any second edge connecting the fault cause node and the faulty component node in the first knowledge graph, the first weight is the ratio of the edge attribute of the second edge connecting the fault cause node and the faulty component node to the sum of the edge attributes of all the second edges connecting the fault cause node.

[0118] In this embodiment, by setting the attributes of the edges in the knowledge graph, it is possible to update the first weight in the first knowledge graph based on the update of the attributes of the edges in the knowledge graph. The update method of the first weight in the first knowledge graph is relatively simple.

[0119] Specifically, before step 101, the following steps are also included:

[0120] Acquire historical fault data, which includes multiple fault information of the same vehicle model within a historical time period. Each fault information includes the corresponding fault phenomenon, fault cause, and faulty component.

[0121] In this embodiment, historical fault data can be obtained from the vehicle maintenance system, which records first-level raw data including vehicle model, fault symptoms, fault causes, and faulty components during vehicle maintenance. Multiple fault information entries for the same vehicle model can be first filtered from the first-level raw data. Then, these entries can be filtered based on historical time periods. Finally, disambiguation and removal processing is performed on the filtered fault information to obtain historical fault data. The removal processing removes meaningless words from the fault information, such as common high-frequency words, stop words, punctuation marks, emoticons, and simplified / traditional character conversions.

[0122] Historical fault data includes multiple fault information for the same vehicle model within a historical time period. For example, for the Haval H6 model, there are corresponding fault information 1, fault information 2, fault information 3, etc. Fault information 1 can include: fault phenomenon - window cannot be closed, fault cause - battery dead, fault component - battery. Fault information 2 can include: fault phenomenon - windshield wipers cannot be used, fault cause - wiper controller damaged, fault component - windshield wipers. Fault information 3 can include: fault phenomenon - window cannot be closed, fault cause - wire breakage, fault component - wire interface.

[0123] Based on historical fault data, determine the weight of the fault cause in its corresponding fault phenomenon.

[0124] In this embodiment, for multiple fault information in the historical fault data, a first triplet can be extracted from each fault information. That is, multiple first triplets can be extracted based on multiple fault information in the historical fault data. The first triplet includes the fault phenomenon and the fault cause. The first triplet is (fault phenomenon, cause, fault cause). For example, the first triplet can be (window cannot be closed, cause, battery dead).

[0125] For the multiple first triplets obtained, the total number of times the same fault phenomenon occurred can be counted. For example, the fault phenomenon of a car window not closing occurred a total of 20 times. Then, for this fault phenomenon, the number of times the corresponding fault cause occurred can be counted. For example, for the fault phenomenon of a car window not closing, the fault cause of a dead battery occurred 2 times, the fault cause of a broken wire occurred 10 times, the fault cause of a damaged controller occurred 7 times, and the fault cause of aging rubber occurred 1 time. Then, based on the ratio of the number of times the fault cause corresponding to the fault phenomenon occurred to the total number of times the same fault phenomenon occurred, the weight of the fault cause in its corresponding fault phenomenon can be determined. For example, if the fault cause of a dead battery occurred 2 times, and the fault phenomenon of a car window not closing occurred a total of 20 times, then the weight of the fault cause of a dead battery in its corresponding fault phenomenon of a car window not closing is the ratio of 2 to 20, which is 0.1.

[0126] After determining the weight of the cause of the failure in its corresponding failure symptom, it can be stored in the RDF (Resource Description Framework) data format to form the attribute relationship between the failure symptom and the cause of the failure. RDF is a markup language used to describe web resources. It is an XML (a subset of Standard Generalized Markup Language) application that processes metadata. Metadata is "data that describes data" or "information that describes information".

[0127] Based on historical fault data, determine the weight of the faulty component in its corresponding fault cause.

[0128] In this embodiment, for multiple fault information in historical fault data, a second triplet can be extracted from each fault information. That is, multiple second triplets can be extracted based on multiple fault information in historical fault data. The second triplet includes the fault cause and the faulty component. The second triplet is (fault cause, component, faulty component). For example, the first triplet can be (battery dead, component, battery).

[0129] For multiple second triplets obtained, the total number of times the same fault cause occurred can be counted. For example, the fault cause of a dead battery occurred a total of 20 times. Then, for each fault cause, the number of times the corresponding faulty component failed can be counted. For example, for the fault cause of a dead battery, the number of times the battery failed is 2. Then, based on the ratio of the number of times the corresponding faulty component failed to fail to fail to the total number of times the same fault cause occurred, the weight of the faulty component in its corresponding fault cause can be determined. For example, if the battery failed 2 times and the fault cause of a dead battery occurred a total of 2 times, then the weight of the battery's faulty component in its corresponding fault cause of a dead battery is the ratio of 2 to 2, which is 1.

[0130] After determining the weight of the faulty component in its corresponding fault cause, it can be stored in RDF data format to form the attribute relationship between the faulty component and the fault cause.

[0131] A first knowledge graph is constructed based on historical fault data, the weight of fault causes in their corresponding fault phenomena, and the weight of faulty components in their corresponding fault causes.

[0132] In this embodiment, a first knowledge graph can be constructed based on multiple first triples and multiple second triples obtained from historical fault data, as well as the weights of fault causes in their corresponding fault phenomena and the weights of faulty components in their corresponding fault causes. The constructed first knowledge graph can refer to... Figure 3 It should be noted that when constructing the first knowledge graph, for the same first triplet or second triplet, there is no need to construct it repeatedly; only a corresponding node and edge need to be constructed.

[0133] In addition, when constructing the first knowledge graph, it is also necessary to obtain the edge attributes of the edges in the first knowledge graph based on the number of times the fault cause corresponding to the fault phenomenon occurs and the number of times the faulty component corresponding to the fault cause fails.

[0134] In this embodiment, the construction of the first knowledge graph requires consideration of fault symptoms, fault causes, and faulty components. Based on this first knowledge graph, the fault causes and faulty components can be quickly located when a vehicle malfunctions, enabling targeted inspection and repair. This helps repair personnel quickly diagnose and efficiently resolve problems, greatly improving work efficiency and thus enhancing the efficiency of vehicle diagnostics. Furthermore, the construction of the first knowledge graph also considers the weight of the fault cause in its corresponding fault symptom and the weight of the faulty component in its corresponding fault cause. This allows for inspection from fault causes and faulty components with higher weights, further improving the efficiency of vehicle diagnostics.

[0135] Specifically, step 102 includes:

[0136] Obtain current fault data, which includes fault cause analysis information and handling process information for the same vehicle model within the current time period.

[0137] In this embodiment, historical fault data can be obtained from the vehicle maintenance system. The vehicle maintenance system records second raw data during vehicle maintenance, including service station number, vehicle model, fault phenomenon, fault cause, faulty component, cause analysis information, and processing information. The first raw data includes multiple current fault information for the same vehicle model. Each current fault information includes fault cause analysis information and processing information. The fault cause analysis information and processing information for the same vehicle model within the current time period can be extracted from the multiple current fault information in the first raw data.

[0138] Extract the cause triplet between the fault phenomenon and the fault cause from the fault cause analysis information.

[0139] In this embodiment, the fault cause analysis information includes an analysis of the fault phenomenon and the fault cause that led to the fault phenomenon. For example, the fault cause analysis information may include: regarding the fault phenomenon that the vehicle window cannot be closed, an inspection of multiple structures related to the window reveals that the fault cause is a dead battery. Semantic analysis can then be performed on the fault cause analysis information to extract the cause triples between the fault phenomenon and the fault cause. A cause triple may include (fault phenomenon, cause, fault cause), such as (window cannot be closed, cause, dead battery), where the cause is the relationship between the entity "window cannot be closed" and the entity "dead battery."

[0140] Of course, after extracting the cause triples, for each fault phenomenon, the frequency of occurrence of the corresponding cause can be counted. For example, for the fault phenomenon that the car window cannot be closed, the cause of the dead battery occurred once. Furthermore, the extracted cause triples can be stored in RDF data format.

[0141] Extract the component triplet between the fault cause and the faulty component from the processing information.

[0142] In this embodiment, the processing information includes records of specific handling of the fault causes. For example, the processing information may include: regarding the fault phenomenon that the vehicle's windows cannot be closed, the fault cause was found to be a dead battery, further investigation revealed that the faulty component was the battery, and then the battery was repaired accordingly. Semantic analysis can then be performed on the processing information to extract component triples between the fault cause and the faulty component. Component triples may include (fault cause, component, faulty component), such as (dead battery, component, battery), where the component represents the relationship between the entity "dead battery" and the entity "battery".

[0143] Of course, after extracting the component triples, for each fault cause, the number of times the corresponding faulty component fails can be counted. For example, for a fault cause of a dead battery, the number of times the battery fails is 1. Furthermore, the extracted component triples can be stored in RDF data format.

[0144] Construct a second knowledge graph based on cause triples and component triples.

[0145] In this embodiment, a second knowledge graph can be constructed based on the obtained cause triples and component triples. Furthermore, during the construction of the second knowledge graph, the edge attributes of the edges in the second knowledge graph can be obtained based on the number of times the fault cause corresponding to the fault phenomenon occurs and the number of times the faulty component corresponding to the fault cause fails. The constructed second knowledge graph can refer to... Figure 4 The second knowledge graph can be displayed in the form of a tree diagram, forming a structure like "car window cannot be closed - battery dead - battery".

[0146] Of course, when constructing the second knowledge graph, it can also be constructed based on the weight of the fault cause in its corresponding fault phenomenon and the weight of the fault component in its corresponding fault cause. The second knowledge graph constructed in this way includes the second weight of each edge.

[0147] In this embodiment, a second knowledge graph is constructed based on the fault cause analysis information and processing information of the same vehicle model within the current time period. That is, the second knowledge graph is constructed based on the fault data within the current time period that is close to the current time, which can ensure the dynamic updating and real-time nature of the first knowledge graph.

[0148] After the first knowledge graph is updated, the methods for using the updated first knowledge graph include:

[0149] Obtain the latest first knowledge graph;

[0150] Receive fault symptom query information input by the user;

[0151] Based on the fault phenomenon query information, the target fault phenomenon node corresponding to the fault phenomenon query information is determined from the latest first knowledge graph, and all target fault cause nodes connected to the target fault phenomenon node through the first edge are determined, and all target fault component nodes connected to the target fault cause node through the second edge are determined.

[0152] The order of the target fault cause nodes is determined based on the first weight of the first side. The larger the first weight, the higher the order of the corresponding target fault cause node.

[0153] Output the fault cause feedback results based on the sorting of the target fault cause nodes, and output the fault component feedback results associated with the fault cause feedback results based on the target fault component nodes.

[0154] Secondly, referring to Figure 5 This illustration shows a schematic diagram of a knowledge graph updating device provided in an embodiment of the present invention. The present invention provides a knowledge graph updating device, comprising:

[0155] The first knowledge graph acquisition module 201 is used to acquire a first knowledge graph, wherein the first knowledge graph is constructed based on historical fault data. The first knowledge graph includes nodes, edges, and a first weight for each edge. The nodes of the first knowledge graph include fault phenomenon nodes, fault cause nodes, and fault component nodes. The edges of the first knowledge graph include a first edge connecting the fault phenomenon node and the fault cause node and a second edge connecting the fault cause node and the fault component node. The first weight of the first edge is used to represent the importance of the fault cause corresponding to the first edge, and the first weight of the second edge is used to represent the importance of the fault component corresponding to the second edge.

[0156] The second knowledge graph construction module 202 is used to construct a second knowledge graph based on the current fault data.

[0157] The update module 203 is used to update the first weight in the first knowledge graph based on the second knowledge graph, and / or expand the nodes and edges in the first knowledge graph.

[0158] Specifically, the second knowledge graph includes nodes and edges. The nodes of the second knowledge graph include fault phenomenon nodes, fault cause nodes, and faulty component nodes; the update module 203 includes:

[0159] The first extraction unit is used to extract multiple first vector expressions from the nodes and edges of the first knowledge graph, wherein the first vector expressions include the corresponding fault phenomena, fault causes and faulty components.

[0160] The second extraction unit is used to extract multiple second vector expressions from the nodes and edges of the second knowledge graph, wherein the second vector expressions include the corresponding fault phenomena, fault causes and faulty components.

[0161] The first determining unit is used to determine the matching degree between the second vector expression and the first vector expression;

[0162] The first update unit is used to update the first weight in the first knowledge graph according to the second vector expression if the matching degree between the second vector expression and the first vector expression is greater than or equal to the matching preset value.

[0163] The second update unit is used to expand the nodes and edges in the first knowledge graph according to the second vector expression if the matching degree between the second vector expression and all the first vector expressions is less than the matching preset value, or to expand the nodes and edges in the first knowledge graph and update the first weight in the first knowledge graph.

[0164] Specifically, the nodes in both the first and second knowledge graphs include fault phenomenon nodes, fault cause nodes, and fault component nodes. Fault phenomenon nodes and fault cause nodes are connected by edges, and fault cause nodes and fault component nodes are connected by edges.

[0165] Update module 203 also includes:

[0166] The third extraction unit is used to extract a first cause vector expression from the fault phenomenon nodes and fault cause nodes connected by edges in the first knowledge graph, wherein the first cause vector expression includes the corresponding fault phenomenon and fault cause.

[0167] The fourth extraction unit is used to extract the second cause vector expression from the fault phenomenon nodes and fault cause nodes connected by edges in the second knowledge graph, wherein the second cause vector expression includes the corresponding fault phenomenon and fault cause.

[0168] The second determining unit is used to determine the degree of matching between the second cause vector expression and the first cause vector expression;

[0169] The third update unit is used to update the first weight in the first knowledge graph according to the second cause vector expression if the matching degree between the second cause vector expression and the first cause vector expression is greater than or equal to the matching preset value.

[0170] The fourth update unit is used to expand the nodes and edges in the first knowledge graph according to the second cause vector expression if the matching degree between the second cause vector expression and all the first cause vector expressions is less than the matching preset value, or to expand the nodes and edges in the first knowledge graph and update the first weight in the first knowledge graph.

[0171] Specifically, update module 203 also includes:

[0172] The fifth extraction unit is used to extract the first component vector expression from the fault cause nodes and fault component nodes connected by edges in the first knowledge graph, wherein the first component vector expression includes the corresponding fault cause and fault component.

[0173] The sixth extraction unit is used to extract the second component vector expression from the fault cause nodes and fault component nodes connected by edges in the second knowledge graph, wherein the second component vector expression includes the corresponding fault cause and fault component;

[0174] The third determining unit is used to determine the matching degree between the second component vector expression and the first component vector expression;

[0175] The fifth update unit is used to update the first weight in the first knowledge graph according to the second component vector expression if the matching degree between the second component vector expression and the first component vector expression is greater than or equal to the matching preset value.

[0176] The sixth update unit is used to expand the nodes and edges in the first knowledge graph according to the second component vector expression if the matching degree between the second component vector expression and all the first component vector expressions is less than the matching preset value, or to expand the nodes and edges in the first knowledge graph and update the first weight in the first knowledge graph.

[0177] Specifically, the edges in both the first and second knowledge graphs have attributes. The edge attributes in the first knowledge graph include the first number, and the edge attributes in the second knowledge graph include the second number.

[0178] Update the first weight in the first knowledge graph according to the second vector expression, including:

[0179] Update the edge attributes of the edge corresponding to the first vector expression based on the edge attributes of the edge corresponding to the second vector expression.

[0180] Update the first weight in the first knowledge graph based on the edge attributes of the edge corresponding to the updated first vector expression.

[0181] Specifically, the knowledge graph updating device also includes:

[0182] The first acquisition unit is used to acquire historical fault data, which includes multiple fault information of the same vehicle model within a historical time period. Each fault information includes the corresponding fault phenomenon, fault cause and fault component.

[0183] The fourth determining unit is used to determine the weight of the cause of the fault in its corresponding fault phenomenon based on historical fault data.

[0184] The fifth is to determine the weight of the faulty component in its corresponding fault cause based on historical fault data;

[0185] The first construction unit is used to construct the first knowledge graph based on historical fault data, the weight of fault causes in their corresponding fault phenomena, and the weight of faulty components in their corresponding fault causes.

[0186] Specifically, the second knowledge graph construction module 202 includes:

[0187] The second acquisition unit is used to acquire current fault data, which includes fault cause analysis information and processing information for the same vehicle model within the current time period.

[0188] The seventh extraction unit is used to extract the cause triplet between the fault phenomenon and the fault cause from the fault cause analysis information;

[0189] The eighth extraction unit is used to extract the component triplet between the fault cause and the faulty component from the processing information;

[0190] The second building unit is used to construct a second knowledge graph based on cause triples and component triples.

[0191] Since the knowledge graph updating device and the knowledge graph updating method share the same technical concept, the device also possesses the beneficial effects of the knowledge graph updating method, and will not be elaborated further here. As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple; relevant details can be found in the descriptions of the method embodiment.

[0192] Based on the same inventive concept, another embodiment of the present invention provides an electronic device, including: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to execute the instructions to implement the knowledge graph update method as described in any of the above embodiments of the present invention.

[0193] Based on the same inventive concept, another embodiment of the present invention provides a readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the knowledge graph update method as described in any of the above embodiments of the present invention.

[0194] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. Those skilled in the art will understand that the embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, the embodiments of the present invention can take the form of entirely hardware embodiments, entirely software embodiments, or embodiments combining software and hardware aspects. Furthermore, the embodiments of the present invention can take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0195] Embodiments of the present invention are described with reference to flowchart illustrations and / or block diagrams of methods, apparatus, and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0196] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0197] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal equipment, causing a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0198] Although preferred embodiments of the present invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present invention.

[0199] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes the aforementioned element.

[0200] The above provides a detailed description of the knowledge graph updating method, apparatus, electronic device, and storage medium provided by the present invention. Specific examples have been used to illustrate the principles and embodiments of the present invention. The description of the embodiments above is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in specific embodiments and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A knowledge graph updating method, characterized in that, include: Obtain a first knowledge graph, wherein the first knowledge graph is constructed based on historical fault data, the first knowledge graph includes nodes, edges and a first weight for each edge, the nodes of the first knowledge graph include fault phenomenon nodes, fault cause nodes and fault component nodes, the edges of the first knowledge graph include a first edge connecting the fault phenomenon node and the fault cause node and a second edge connecting the fault cause node and the fault component node, the first weight of the first edge is used to represent the importance of the fault cause corresponding to the first edge, and the first weight of the second edge is used to represent the importance of the fault component corresponding to the second edge; Based on the current fault data, a second knowledge graph is constructed; the second knowledge graph includes nodes and edges, and the nodes of the second knowledge graph include fault phenomenon nodes, fault cause nodes, and fault component nodes; Based on the second knowledge graph, update the first weight in the first knowledge graph, and / or expand the nodes and edges in the first knowledge graph, including: Multiple first vector expressions are extracted from the nodes and edges of the first knowledge graph, wherein the first vector expression includes the corresponding fault phenomenon, fault cause and fault component; Multiple second vector expressions are extracted from the nodes and edges of the second knowledge graph, wherein the second vector expressions include the corresponding fault phenomena, fault causes and faulty components; Determine the degree of matching between the second vector expression and the first vector expression; If the matching degree between the second vector expression and the first vector expression is greater than or equal to the preset matching value, then the first weight in the first knowledge graph is updated according to the second vector expression; If the matching degree between the second vector expression and all the first vector expressions is less than the preset matching value, then the nodes and edges in the first knowledge graph are expanded according to the second vector expression, or the nodes and edges in the first knowledge graph are expanded and the first weight in the first knowledge graph is updated.

2. The method according to claim 1, characterized in that, The step of updating the first weights between nodes in the first knowledge graph and / or expanding the nodes and edges in the first knowledge graph according to the second knowledge graph further includes: A first cause vector expression is extracted from the fault phenomenon nodes and fault cause nodes connected by edges in the first knowledge graph, wherein the first cause vector expression includes the corresponding fault phenomenon and fault cause. The second cause vector expression is extracted from the fault phenomenon nodes and fault cause nodes connected by edges in the second knowledge graph, wherein the second cause vector expression includes the corresponding fault phenomenon and fault cause. Determine the degree of matching between the second cause vector expression and the first cause vector expression; If the matching degree between the second cause vector expression and the first cause vector expression is greater than or equal to the matching preset value, then the first weight in the first knowledge graph is updated according to the second cause vector expression; If the matching degree between the second cause vector expression and all the first cause vector expressions is less than the matching preset value, then the nodes and edges in the first knowledge graph are expanded according to the second cause vector expression, or the nodes and edges in the first knowledge graph are expanded and the first weight in the first knowledge graph is updated.

3. The method according to claim 1, characterized in that, Based on the second knowledge graph, updating the first weights between nodes in the first knowledge graph, and / or expanding the nodes and edges in the first knowledge graph, further includes: The first component vector expression is extracted from the fault cause nodes and fault component nodes connected by edges in the first knowledge graph, wherein the first component vector expression includes the corresponding fault cause and fault component. The second component vector expression is extracted from the fault cause nodes and fault component nodes connected by edges in the second knowledge graph, wherein the second component vector expression includes the corresponding fault cause and fault component; Determine the matching degree between the second component vector expression and the first component vector expression; If the matching degree between the second component vector expression and the first component vector expression is greater than or equal to the matching preset value, then the first weight in the first knowledge graph is updated according to the second component vector expression; If the matching degree between the second component vector expression and all the first component vector expressions is less than a preset matching value, then the nodes and edges in the first knowledge graph are expanded according to the second component vector expression, or the nodes and edges in the first knowledge graph are expanded and the first weight in the first knowledge graph is updated.

4. The method according to claim 1, characterized in that, The edges in both the first and second knowledge graphs have attributes. The edge attributes in the first knowledge graph include the first number, and the edge attributes in the second knowledge graph include the second number. The step of updating the first weight in the first knowledge graph according to the second vector expression includes: Update the edge attributes of the edge corresponding to the first vector expression based on the edge attributes of the edge corresponding to the second vector expression; Update the first weight in the first knowledge graph based on the edge attributes of the edge corresponding to the updated first vector expression.

5. The method according to claim 1, characterized in that, Before the step of obtaining the first knowledge graph, the method further includes: Acquire historical fault data, wherein the historical fault data includes multiple fault information of the same vehicle model within a historical time period, and each fault information includes the corresponding fault phenomenon, fault cause and fault component; Based on the historical fault data, determine the weight of the fault cause in its corresponding fault phenomenon; Based on the historical fault data, the weight of the faulty component in its corresponding fault cause is determined; A first knowledge graph is constructed based on the historical fault data, the weight of the fault cause in its corresponding fault phenomenon, and the weight of the faulty component in its corresponding fault cause.

6. The method according to claim 1, characterized in that, The construction of the second knowledge graph based on the current fault data includes: Acquire current fault data, wherein the current fault data includes fault cause analysis information and processing information for the same vehicle model within the current time period; Extract the cause triplet between the fault phenomenon and the fault cause from the fault cause analysis information; Extract the component triplet between the fault cause and the faulty component from the processing information; A second knowledge graph is constructed based on the cause triples and the component triples.

7. A knowledge graph updating device, characterized in that, include: The first knowledge graph acquisition module is used to acquire a first knowledge graph, wherein the first knowledge graph is constructed based on historical fault data. The first knowledge graph includes nodes, edges, and a first weight for each edge. The nodes of the first knowledge graph include fault phenomenon nodes, fault cause nodes, and fault component nodes. The edges of the first knowledge graph include a first edge connecting the fault phenomenon node and the fault cause node and a second edge connecting the fault cause node and the fault component node. The first weight of the first edge is used to represent the importance of the fault cause corresponding to the first edge, and the first weight of the second edge is used to represent the importance of the fault component corresponding to the second edge. The second knowledge graph construction module is used to construct a second knowledge graph based on the current fault data; the second knowledge graph includes nodes and edges, and the nodes of the second knowledge graph include fault phenomenon nodes, fault cause nodes, and fault component nodes; The update module is used to update the first weight in the first knowledge graph according to the second knowledge graph, and / or expand the nodes and edges in the first knowledge graph; The update module includes: The first extraction unit is used to extract multiple first vector expressions from the nodes and edges of the first knowledge graph, wherein the first vector expression includes the corresponding fault phenomenon, fault cause and fault component. The second extraction unit extracts multiple second vector expressions from the nodes and edges of the second knowledge graph, wherein the second vector expressions include the corresponding fault phenomena, fault causes and faulty components. The first determining unit is used to determine the matching degree between the second vector expression and the first vector expression; The first update unit is used to update the first weight in the first knowledge graph according to the second vector expression if the matching degree between the second vector expression and the first vector expression is greater than or equal to a preset matching value. The second update unit is configured to, if the matching degree between the second vector expression and all the first vector expressions is less than a preset matching value, expand the nodes and edges in the first knowledge graph according to the second vector expression, or expand the nodes and edges in the first knowledge graph and update the first weight in the first knowledge graph.

8. An electronic device, characterized in that, include: processor; Memory used to store the processor's executable instructions; The processor is configured to execute the instructions to implement the knowledge graph update method as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the knowledge graph update method according to any one of claims 1 to 6.

Images