An automobile part packaging design case knowledge graph construction method and system
By evaluating the strong update frequency and historical query frequency of change cases, the node relationships in the knowledge graph are dynamically updated, solving the timeliness problem of knowledge graphs in automotive parts packaging design and improving the accuracy of the knowledge graph and the utilization efficiency of storage resources.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- YINGLING TECHNOLOGY (BEIJING) CO LTD
- Filing Date
- 2026-03-02
- Publication Date
- 2026-06-05
Smart Images
Figure CN122154680A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of knowledge graph construction technology, specifically to a knowledge graph construction method and system for an automotive parts packaging design case. Background Technology
[0002] Automotive parts packaging design refers to designing packaging solutions for various components such as engines, transmissions, and brake pads, taking into account their size, weight, and fragility, to provide protection, transportation, and storage functions. Traditional design relies on engineers' experience, resulting in low efficiency and inconsistent solutions. Therefore, it is necessary to construct a knowledge graph to structure and store scattered design cases, standards, specifications, and material parameters, enabling functions such as intelligent recommendation, conflict detection, and solution optimization.
[0003] The existing knowledge graph construction process includes: domain ontology design and modeling, defining core entities and their relationships; multi-source heterogeneous data collection and preprocessing, integrating case libraries, technical documents, standards and other data; ontology-based knowledge extraction, identifying entities, extracting relationships and supplementing attributes; knowledge fusion and quality verification, ensuring knowledge consistency; knowledge storage and graph construction, forming a complete knowledge graph.
[0004] However, the relationships between various elements in automotive parts packaging design are constantly changing. With the innovative application of environmentally friendly materials, fluctuations in logistics market prices, and continuous updates to relevant regulations and policies, the original "parts-packaging material" matching relationship may become invalid. Most existing knowledge graphs are based on static relationship assumptions and lack timeliness management, resulting in the retention of a large amount of outdated and invalid information in the knowledge graphs. Summary of the Invention
[0005] In light of the above, it is necessary to provide a method and system for constructing a knowledge graph for automotive parts packaging design cases, which improves the accuracy and timeliness of the knowledge graph compared to traditional methods. In a first aspect, embodiments of this application provide a method for constructing a knowledge graph for automotive parts packaging design cases, the method comprising the following steps: Collect change cases of automotive parts packaging design, each change case includes a timestamp, multiple keyword groups, and multiple triples; If a new change case is received at the current moment, execute the following knowledge graph maintenance process: By analyzing the frequency of occurrence of each keyword group in each new change case within the nearest neighbor time period at the current moment, and the frequency of triggering strong update operations in historical change cases, the strong update degree of each new change case is obtained. Then, the strong update degree of each triple in each new change case is obtained, and corrected using the chain dependency relationships in the knowledge graph to obtain the final strong update degree of each triple in each new change case. The relationships between nodes in the knowledge graph are then updated based on the final strong update degree. Add strong update events to the historical records of each node that is updating node relationships. By using the total number of times each node is queried, accessed, and referenced in each historical strong update interval in the knowledge graph, as well as the time difference between the initial time of each historical strong update interval and the current time, the availability of each node in each historical strong update interval is obtained. Then, combined with the length of each historical strong update interval, the time interval for periodically cleaning up the relationships of each node is obtained, and the relationships of each node are periodically cleaned up.
[0006] In one embodiment, the process of obtaining the strong update degree of each new change case is as follows: Calculate the product of the inverse proportional mapping result of the occurrence frequency and the frequency; The strong update degree of each new change case is obtained by multiplying the products of all keyword groups in each new change case.
[0007] In one embodiment, the strong update degree of each new change case is the maximum value among the products corresponding to all keyword groups in each new change case.
[0008] In one embodiment, the process of obtaining the strong update degree of each triple in each new change case is as follows: Calculate the shortest path length in the knowledge graph for the head entity, tail entity, and keyword group corresponding to the maximum value of each triple in each new change case, and select the minimum value among the shortest path lengths; The product of the inverse proportional mapping result of the minimum value and the strong update degree of each new change case is used as the strong update degree of each triple in each new change case.
[0009] In one embodiment, the process of obtaining the final strong update degree of each triple in each new change case is as follows: Obtain the upstream dependency path of any triple in the knowledge graph for each new change case, and set the maximum backtracking depth; The strong update degree of each triple on the upstream dependency path of any triple is multiplied by the corresponding preset decay degree, and compared with the strong update degree of any triple itself. The maximum value is selected as the final strong update degree of any triple.
[0010] In one embodiment, updating the relationships between nodes in the knowledge graph includes: If the final strong update degree of each triple in each new change case is greater than the preset strong update judgment threshold, it is judged as a strong update. The original relationship of the head entity of each triple in each new change case in the knowledge graph is marked as a historical failure state and the failure time is recorded. At the same time, new connection relationships are established based on each triple in each new change case. Otherwise, the relationships that originally existed in the knowledge graph for the head entities of each triple in each new change case will be retained in the knowledge graph, and alternative relationships will be established for incremental expansion based on each triple in each new change case.
[0011] In one embodiment, the process of obtaining the availability of each historical strong update time interval of each node is as follows: The product of the normalized value of the total number of times and the inverse proportional mapping result of the time difference is used as the availability of each historical strong update time interval of each node.
[0012] In one embodiment, the process of obtaining the time interval for periodic cleaning is as follows: The product of the reciprocal of each historical strong update time interval of each node and its availability is denoted as the active product. The cumulative value of the active product of all historical strong update time intervals of each node is calculated. The time interval for periodic cleaning is inversely proportional to the accumulated value.
[0013] In one embodiment, the periodic cleaning of the relationships between nodes includes: The time interval for periodically cleaning up the relationships of each node is used as the cycle, and the trigger time for the next cleaning of the relationships of each node is set. When the task is triggered, the call frequency of each relationship of each node is counted, and the relationships whose call frequency is lower than the preset retention frequency threshold are cleaned up.
[0014] Secondly, embodiments of this application also provide a knowledge graph construction system for automotive parts packaging design cases, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the knowledge graph construction method for automotive parts packaging design cases described above.
[0015] This application has at least the following beneficial effects: Upon receiving a new change case, this application initiates a maintenance process to ensure that the knowledge graph can reflect the latest design changes in a timely manner. By analyzing the frequency of keyword groups triggering strong update operations in historical change cases, the strong update indication capability of each keyword group is quantified. Combined with the frequency of occurrence of keyword groups in the nearest neighbor time period at the current moment, "low-frequency, high-impact" change events are captured, enabling accurate identification of the importance and scope of impact of new change cases. Furthermore, by combining the chain dependency relationships in the knowledge graph, the strong updateability of triples is comprehensively evaluated. Based on the final strong updateability measurement results, the relationships between nodes in the knowledge graph are updated in a timely manner to ensure that the relationships in the knowledge graph always reflect the latest business status, avoid interference from outdated information on the knowledge graph, and improve the accuracy and reliability of the knowledge graph. Furthermore, by comprehensively considering data timeliness and business activity, the availability of each node's historical strong update intervals is measured, and the interval for periodically cleaning up the relationships between each node is dynamically adjusted. This ensures the scientific and rational nature of the cleaning strategy. Through periodic cleaning, redundant relationships are removed, storage resources are optimized, and the timeliness of the knowledge graph is improved. Attached Figure Description
[0016] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0017] Figure 1 A flowchart illustrating the steps of a knowledge graph construction method for an automotive parts packaging design case, provided in one embodiment of this application; Figure 2 This is a schematic diagram illustrating the process of obtaining the final strong update degree. Detailed Implementation
[0018] In the description of the embodiments in this application, the words "exemplary," "or," and "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the words "exemplary," "or," and "for example" is intended to present the relevant concepts in a specific manner.
[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of this application.
[0020] It should also be noted that the terms "first" and "second" in this application are used to distinguish similar objects, rather than to describe a specific order or sequence.
[0021] The following, in conjunction with the accompanying drawings, details the specific scheme of the knowledge graph construction method and system for an automotive parts packaging design case provided in this application.
[0022] Please see Figure 1 The diagram illustrates a flowchart of a knowledge graph construction method for an automotive parts packaging design case according to an embodiment of this application. The method includes the following steps: Step 1: Collect change cases of automotive parts packaging design. Each change case includes a timestamp, multiple keyword groups, and multiple triples.
[0023] We collect change cases for automotive parts packaging design from multi-source heterogeneous data, specifically covering packaging design case libraries, technical documents, industry standards, and supplier technical materials. For structured data, we directly acquire it through API interfaces with enterprise ERP / PLM systems; for unstructured text data, we automatically extract entities, relationships, and keyword phrases reflecting business dynamics using natural language processing technology. All change cases are assigned a timestamp attribute to create a time index.
[0024] The acquired change cases underwent data cleaning, format standardization, and deduplication and noise reduction to unify the multi-source data into a structured format, forming a time-series case dataset containing triples, keyword phrases, and time attributes. The time-series case dataset is shown in Table 1. Table 1. Temporal Case Dataset Step 2: If a new change case is received at the current moment, execute the following knowledge graph maintenance process.
[0025] Step 2.1: Obtain the strong update degree of each new change case by the frequency of occurrence of each keyword group in the nearest neighbor time period at the current moment and the frequency of triggering strong update operations in historical change cases. Then, obtain the strong update degree of each triple in each new change case and correct it using the chain dependency relationship in the knowledge graph to obtain the final strong update degree of each triple in each new change case. Update the relationship of nodes in the knowledge graph using the final strong update degree.
[0026] This application primarily targets sudden changes that lack prior rules, occur infrequently but have a wide impact, such as unexpected supply disruptions from specific suppliers or bans on niche materials.
[0027] (1) Obtain the strong update degree of each new change case by the frequency of occurrence of each keyword group in the current time period in the current time and the frequency of triggering strong update operation in the historical change cases.
[0028] In the field of automotive parts packaging design, certain phrases clearly indicate strong updates. For example, regulatory phrases include: "environmental regulations updated," "banned materials," "standard upgrades," and "policy changes"; technology phrases include: "material phase-out," "technological breakthroughs," "process changes," and "product iterations"; and constraint phrases include: "sole supplier," "mandatory requirements," and "exclusivity." In historical change cases, these phrases typically correspond to strong update operations that require breaking old relationships and establishing new ones. In short, strong updates refer to operations that require blocking, removing, or marking existing entity relationships in the knowledge graph as invalid; weak updates refer to changes such as the introduction of new solutions or the addition of alternative materials, which only require incremental expansion by adding new entities or relationships to the knowledge graph without disrupting the existing structure.
[0029] When the aforementioned phrases appear, they often trigger a disconnection operation in the knowledge graph. This causal relationship forms a statistical regularity in historical data. The ratio of "the number of times any phrase triggers a strong update operation to the total number of times that phrase appears" is used as an indicator of strong updateability. The larger the ratio, the stronger the reliability of that phrase as a strong update trigger condition. This statistical method based on the historical frequency of strong update operations can objectively quantify the strong update indicative ability of phrases, avoid subjective judgment bias, and provide a measurable basis for subsequently measuring the strong updateability of change cases.
[0030] Based on the above analysis, the frequency with which each keyword group in each new change case triggers a strong update operation in historical change cases is used to obtain the strong update degree of each keyword group in each new change case. This degree is used to measure the strong updateability of each keyword group in each new change case, and the expression is: In the formula, This indicates the strong update degree of the i-th keyword phrase in the new change case A; This indicates the number of times the i-th keyword phrase in the new change case A triggers a strong update operation in the historical change cases; This indicates the number of times the i-th keyword phrase in new change case A appears in historical change cases. It should be added that: if... If the value is 0, calculate first. The sum of 1 and 1 is then used for further calculations.
[0031] Furthermore, the automotive parts packaging design field exhibits a "low-frequency, high-impact" characteristic. Major update events such as policy and regulatory changes, and technological breakthroughs, occur infrequently due to long cycles and high costs, but their impact is far-reaching once they occur. Therefore, when a major update phrase, such as "standard upgrade," appears infrequently recently, it is more likely to indicate a genuine major update event. This is because frequently occurring phrases are more likely to be routine information supplements or false alarms. This combination of "stronger update nature and lower recent frequency" precisely captures the scarcity and importance of major update events, amplifying the weight of low-frequency events through an inverse relationship, making the judgment more consistent with the actual business patterns in the automotive parts packaging design field.
[0032] Based on the above analysis, the strong updateability of each new change case is obtained, which is used to measure the strong updateability of each new change case. The expression is: In the formula, This indicates the strong update degree of the new change case A; max{} indicates the operation of selecting the maximum value; exp() represents the exponential function with the natural constant as the base, used to perform inverse proportional mapping on the data; This represents the number of times the i-th keyword phrase appears in the nearest neighbor time period of the current moment in the newly changed case A; This indicates the strong update degree of the i-th keyword group in the new change case A; i represents the sequence number of the keyword group in the new change case A. This indicates the total number of keyword phrases in the new change case A.
[0033] In this embodiment, the nearest neighbor time period refers to the past month. The value of the nearest neighbor time period can be determined by the implementer according to the actual situation. This application does not impose any special restrictions.
[0034] It should be noted that the greater the calculated strong update degree of the new change case A, the more likely the new change case A is to be a strong update event.
[0035] (2) Obtain the strong update degree of each triple in each new change case, and use the chain dependency relationship in the knowledge graph to correct it, so as to obtain the final strong update degree of each triple in each new change case.
[0036] Furthermore, semantic distance can reflect the attenuation pattern of the impact of newly changed cases. The closer the semantic distance between the triple and the strongly updated phrase, the stronger the strong update property of the triple.
[0037] In knowledge graphs, semantic distance between entities, such as the shortest path length, represents the tightness of the concept association. When an entire change case is judged to be a strong update, the strong update phrase is the core trigger point for the change. The closer a triple is to the strong update phrase, the greater the likelihood that it will be directly affected and the higher the probability that it needs a strong update. For example, when the phrase "environmental regulations updated" appears, the nearest triple "a certain material - banned" will definitely need a strong update, while the more distant triple "a certain packaging scheme - cost" may not be affected. This initial judgment based on semantic distance can quickly filter out the triples most likely to be affected, improving the efficiency of subsequent judgments.
[0038] Based on the above analysis, the strong update degree of each triple in each new change case is obtained to measure the strong updateability of each triple in each new change case. The specific process is as follows: Calculate the shortest path length in the knowledge graph for the head entity, tail entity, and keyword group corresponding to the maximum value of each triple in each new change case, and select the minimum value among the shortest path lengths; The product of the inverse proportional mapping result of the minimum value and the strong update degree of each new change case is used as the strong update degree of each triple in each new change case, which is used to measure the strong updateability of each triple in each new change case.
[0039] It should be added that if the keyword group does not initially exist in the knowledge graph, the keyword group is inserted into the knowledge graph as an event node, and an initial edge is established between the node corresponding to the keyword group and the automotive part entity node based on semantic similarity.
[0040] In this embodiment, the opposite of the minimum value is used as the exponent of an exponential function with the natural constant as the base, and the calculation result of the calculated exponential function is used as the inverse proportional mapping result of the minimum value.
[0041] Furthermore, entity relationships in knowledge graphs exhibit significant transmission characteristics, especially in the field of automotive parts packaging design, where a strict "upstream determines downstream" dependency path exists. For example, a change in the upstream "packaging materials," such as the obsolescence of materials, inevitably leads to the failure of the midstream "packaging solution," which in turn affects the downstream "logistics cost" calculation. Therefore, the determination of strong updability ultimately requires the introduction of an inheritance mechanism: downstream nodes should inherit the strong updability attribute of upstream nodes, but this influence should gradually decrease as the dependency path extends.
[0042] Based on the above analysis, the strong update degree of each triple in each new change case is corrected using the chain dependency relationship in the knowledge graph, resulting in the final strong update degree of each triple in each new change case. This final strong update degree is used to measure the final strong update degree of each triple in each new change case. The specific process is as follows: Obtain the upstream dependency path of any triple in the knowledge graph for each new change case, and set the maximum backtracking depth; The strong update degree of each triple on the upstream dependency path of any triple is multiplied by the corresponding preset decay degree, and compared with the strong update degree of any triple itself. The maximum value is selected as the final strong update degree of any triple, which is used to measure the final strong update performance of any triple.
[0043] The expression for the final strong update degree of each triple in each new change case is: In the formula, This indicates the triplet in the new change case A. The final strong update degree; This indicates the maximum value operation, which selects the triples in the new change case A. The most influential path source; This indicates the triplet in the new change case A. High update frequency; This indicates the triplet in the new change case A. The direct upstream dependency triple; This indicates the triplet in the new change case A. The algorithm backtracks up to the upstream dependency triplet at level k; α represents a preset decay factor, ranging from 0.5 to 0.8, used to simulate the characteristic that the update impact weakens as the dependency path length increases; k represents the maximum backtracking depth, used to prevent infinite loops caused by knowledge graph loops and to control computational load. The final flowchart for obtaining the strong update degree is shown below. Figure 2 As shown. Among them, Characterization The preset attenuation level; Characterization The preset attenuation level.
[0044] In this embodiment, the value of α is 0.5.
[0045] In this embodiment, the value of k is 3. The value of k is preset by the user and the implementer can set it according to the actual situation. This application does not impose any special restrictions.
[0046] (3) The relationships between nodes in the knowledge graph are updated using the final strong update degree.
[0047] If the final strong update degree of each triple in each new change case is greater than the preset strong update judgment threshold, it is judged as a strong update. The original relationship of the head entity of each triple in each new change case in the knowledge graph is marked as a historical invalid state and the invalidation time is recorded. At the same time, new connection relationships are established based on each triple in each new change case. It should be added that if the head entity of each triple in each new change case does not originally have a relationship in the knowledge graph, the operation of "marking as a historical invalid state and recording invalidation time" is skipped, and new connection relationships are established only based on each triple in each new change case. If the final strong update degree of each triple in each new change case is less than or equal to the preset strong update judgment threshold, it is judged as a weak update. The original relationship of the head entity of each triple in each new change case in the knowledge graph is retained in the knowledge graph, and alternative relationships are established for incremental expansion based on each triple in each new change case.
[0048] In this embodiment, the preset strong update judgment threshold is set to 0.7, and the preset strong update judgment threshold is calculated through experimental data.
[0049] Step 2.2: Add strong update events to the historical records of each node that is updating node relationships. By using the total number of times each node is queried, accessed and referenced in each historical strong update interval in the knowledge graph, as well as the time difference between the initial time and the current time of each historical strong update interval, obtain the availability of each node in each historical strong update interval. Then, combined with the length of each historical strong update interval, obtain the time interval for periodically cleaning up the relationships of each node, and then perform periodic cleaning of the relationships of each node.
[0050] (1) Add strong update events to the history of each node that performs node relationship updates. The availability of each node in each historical strong update time interval is obtained by the total number of times each node is queried, accessed and referenced in each historical strong update time interval in the knowledge graph, and the time difference between the initial time and the current time of each historical strong update time interval.
[0051] To ensure the timeliness of the knowledge graph while preventing excessive memory load, regular relationship cleanup and maintenance of each node in the knowledge graph are necessary. The maintenance cycle should not be fixed, but dynamically adjusted according to the historical update frequency of each node: for frequently updated and active nodes, the maintenance cycle should be shortened to quickly remove invalid relationships; for stable and less frequently used nodes, the cycle can be extended to save computing power.
[0052] Extract the time sequence of all strong update events that have occurred in the history of each node. Initially, these events can be manually labeled. Subsequently, strong update events are added to the historical records of each node that is updating its node relationships, and the time sequence is updated accordingly. The time span between two temporally adjacent strong update events is defined as the historical strong update interval. The historical strong update interval reflects the "lifespan" of each node between two major changes.
[0053] The reference value of historical strong update intervals is jointly determined by "data timeliness" and "business activity". The closer the historical strong update interval is to the current moment, the better it reflects the current update rhythm; and the more times each node in the knowledge graph is queried, accessed and referenced within each historical strong update interval, the more important the stability of each node is to the business, and the more meaningful the duration of the historical strong update interval of each node is.
[0054] Based on the above analysis, the availability of each node in the knowledge graph for each historical strong update interval is obtained by using the total number of queries, accesses, and references within each historical strong update interval, as well as the time difference between the initial time and the current time of each historical strong update interval. This availability is used to measure the usability of each node for each historical strong update interval. The specific process is as follows: Obtain the normalized value of the total number of queries, accesses and references for each node within each historical strong update interval; obtain the inverse proportional mapping result of the time difference between the initial time and the current time of each historical strong update interval for each node, and use the product of the normalized value and the inverse proportional mapping result as the availability of each node within each historical strong update interval.
[0055] In this embodiment, the formula The calculation result is used as the inverse proportional mapping result of the time difference between the initial time and the current time of each historical strong update time interval, where f() represents the normalization function; Indicates the current moment. This represents the initial time of the *a*th historical strong update interval. The normalization function is the Min-Max normalization function, which is a well-known technique and will not be elaborated upon in this application. The time unit is days; adding 1 is to avoid a denominator of 0, and the unit of 1 is also days.
[0056] Optionally, during the normalization process based on the Min-Max normalization function, if the maximum value equals the minimum value, the normalization value is set to 1.
[0057] (2) By combining the availability of each historical strong update time interval of each node with the length of each historical strong update time interval, the time interval for periodically cleaning up the relationship of each node is obtained.
[0058] By combining the availability of each node's historical strong update interval with the length of each historical strong update interval, the time interval for periodically cleaning up the relationships of each node is obtained, expressed as: In the formula, This indicates the time interval for periodically cleaning up the relationships of a single node; Indicates the preset cleanup interval; N represents the total number of historical strong update intervals; This represents the length of the i-th historical strong update time interval for a single node; This represents the availability of the i-th historical strong update interval for a single node.
[0059] In this embodiment, The value is 365 days. The value is preset by the user, and the implementer can set it according to the actual situation. This application does not impose any special restrictions.
[0060] It should be noted that: This index reflects the current activity level of a single node. The shorter the historical strong update interval, the more frequently a single node updates. A higher recent activity level for a single node indicates higher availability. Increase the size of the node, thereby shortening the time interval for periodically cleaning up the relationships between active nodes; when a single node is stable for a long period of time, This reduces the time interval for periodically cleaning up the relationships between stable nodes.
[0061] It should be added that: if each node has no historical strong update records, or the availability of all historical strong update time intervals is 0, the time interval for periodically cleaning up the relationships of each node will be assigned to 30 days. The calculation result is a decimal, rounded down to the nearest integer.
[0062] (3) Periodically clean up the relationships of each node according to the time interval for periodically cleaning up the relationships of each node.
[0063] The time interval for periodically cleaning up the relationships between nodes is defined as the cycle, and a trigger time is set for the next cleanup of the relationships between nodes. When the task is triggered, the call frequency of each relationship of each node is counted, and relationships with a call frequency lower than a preset retention frequency threshold are cleaned up as redundant relationships. Here, the call frequency refers to the total number of times each relationship of each node is queried, accessed, and referenced within the period of periodic cleanup of the relationships between nodes.
[0064] In this embodiment, the preset retention frequency threshold is set to 5 times / month, and the preset retention frequency threshold is calculated from experimental data.
[0065] Based on the same inventive concept as the above method, this application embodiment also provides a knowledge graph construction system for automotive parts packaging design cases, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-described methods for constructing a knowledge graph for automotive parts packaging design cases.
[0066] In summary, this application initiates a maintenance process upon receiving new change cases to ensure that the knowledge graph can reflect the latest design changes in a timely manner. By analyzing the frequency of keyword groups triggering strong update operations in historical change cases, the strong update indication capability of each keyword group is quantified. Combined with the frequency of occurrence of keyword groups in the nearest neighbor time period at the current moment, "low-frequency, high-impact" change events are captured, enabling accurate identification of the importance and scope of impact of new change cases. Furthermore, by combining the chain dependency relationships in the knowledge graph, the strong updateability of triples is comprehensively evaluated. Based on the final strong updateability measurement results, the relationships between nodes in the knowledge graph are updated in a timely manner to ensure that the relationships in the knowledge graph always reflect the latest business status, avoid interference from outdated information on the knowledge graph, and improve the accuracy and reliability of the knowledge graph. Furthermore, by comprehensively considering data timeliness and business activity, the availability of each node's historical strong update intervals is measured, and the interval for periodically cleaning up the relationships between each node is dynamically adjusted. This ensures the scientific and rational nature of the cleaning strategy. Through periodic cleaning, redundant relationships are removed, storage resources are optimized, and the timeliness of the knowledge graph is improved.
[0067] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than that shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in a different order than disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. Each block in a block diagram and / or flowchart, and combinations of blocks in a block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.
[0068] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from its essential characteristics. Therefore, the embodiments described above should be considered exemplary and non-limiting in all respects.
Claims
1. A method for constructing a knowledge graph for automotive parts packaging design cases, characterized in that, The method includes the following steps: Collect change cases of automotive parts packaging design, each change case includes a timestamp, multiple keyword groups, and multiple triples; If a new change case is received at the current moment, execute the following knowledge graph maintenance process: By analyzing the frequency of occurrence of each keyword group in each new change case within the nearest neighbor time period at the current moment, and the frequency of triggering strong update operations in historical change cases, the strong update degree of each new change case is obtained. Then, the strong update degree of each triple in each new change case is obtained, and corrected using the chain dependency relationships in the knowledge graph to obtain the final strong update degree of each triple in each new change case. The relationships between nodes in the knowledge graph are then updated based on the final strong update degree. Add strong update events to the historical records of each node that is updating node relationships. By using the total number of times each node is queried, accessed, and referenced in each historical strong update interval in the knowledge graph, as well as the time difference between the initial time of each historical strong update interval and the current time, the availability of each node in each historical strong update interval is obtained. Then, combined with the length of each historical strong update interval, the time interval for periodically cleaning up the relationships of each node is obtained, and the relationships of each node are periodically cleaned up.
2. The knowledge graph construction method for an automotive parts packaging design case as described in claim 1, characterized in that, The process for obtaining the strong update degree of each new change case is as follows: Calculate the product of the inverse proportional mapping result of the occurrence frequency and the frequency; The strong update degree of each new change case is obtained by multiplying the products of all keyword groups in each new change case.
3. The knowledge graph construction method for an automotive parts packaging design case as described in claim 2, characterized in that, The strong update degree of each new change case is the maximum value among the products corresponding to all keyword groups in each new change case.
4. The knowledge graph construction method for an automotive parts packaging design case as described in claim 3, characterized in that, The process for obtaining the strong update degree of each triple in each of the new change cases is as follows: Calculate the shortest path length in the knowledge graph for the head entity, tail entity, and keyword group corresponding to the maximum value of each triple in each new change case, and select the minimum value among the shortest path lengths; The product of the inverse proportional mapping result of the minimum value and the strong update degree of each new change case is used as the strong update degree of each triple in each new change case.
5. The knowledge graph construction method for an automotive parts packaging design case as described in claim 1, characterized in that, The process for obtaining the final strong update degree of each triple in each of the new change cases is as follows: Obtain the upstream dependency path of any triple in the knowledge graph for each new change case, and set the maximum backtracking depth; The strong update degree of each triple on the upstream dependency path of any triple is multiplied by the corresponding preset decay degree, and compared with the strong update degree of any triple itself. The maximum value is selected as the final strong update degree of any triple.
6. The knowledge graph construction method for an automotive parts packaging design case as described in claim 1, characterized in that, The updating of the relationships between nodes in the knowledge graph includes: If the final strong update degree of each triple in each new change case is greater than the preset strong update judgment threshold, it is judged as a strong update. The original relationship of the head entity of each triple in each new change case in the knowledge graph is marked as a historical failure state and the failure time is recorded. At the same time, a new connection relationship is established based on each triple in each new change case. Otherwise, it is judged as a weak update. The original relationship of the head entity of each triple in each new change case in the knowledge graph is retained in the knowledge graph. Based on each triple in each new change case, alternative relationships are established for incremental expansion.
7. The knowledge graph construction method for an automotive parts packaging design case as described in claim 1, characterized in that, The process for obtaining the availability of each historical strong update time interval of each node is as follows: The product of the normalized value of the total number of times and the inverse proportional mapping result of the time difference is used as the availability of each historical strong update time interval of each node.
8. The knowledge graph construction method for an automotive parts packaging design case as described in claim 7, characterized in that, The process for obtaining the time interval for the periodic cleanup is as follows: The product of the reciprocal of each historical strong update time interval of each node and its availability is denoted as the active product. The cumulative value of the active product of all historical strong update time intervals of each node is calculated. The time interval for periodic cleaning is inversely proportional to the accumulated value.
9. The knowledge graph construction method for an automotive parts packaging design case as described in claim 1, characterized in that, The periodic cleanup of the relationships between nodes includes: The time interval for periodically cleaning up the relationships of each node is used as the cycle, and the trigger time for the next cleaning of the relationships of each node is set. When the task is triggered, the call frequency of each relationship of each node is counted, and the relationships whose call frequency is lower than the preset retention frequency threshold are cleaned up.
10. A knowledge graph construction system for automotive parts packaging design cases, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the knowledge graph construction method for an automotive parts packaging design case as described in any one of claims 1-9.