Cross-domain data fusion and dynamic knowledge graph construction system for industry chain analysis

Abstract

The application relates to the technical field of knowledge graphs, in particular to a cross-domain data fusion and dynamic knowledge graph construction system for industry chain analysis, which comprises a data processing module, a graph construction module, an analysis module, a robustness determination module, an update adjustment module and a trigger adjustment module. The data processing module comprises a collection unit for collecting multi-source heterogeneous data of multi-field data sources. The graph construction module comprises a construction unit for constructing a knowledge graph according to the cross-domain aligned data. The analysis module is used for analyzing an industry chain according to the incremental knowledge graph to obtain an analysis result. The robustness determination module is used for determining whether the robustness of the multi-source heterogeneous data meets the requirements according to a multi-source data fusion representation value. The update adjustment module is used for determining an attenuation coefficient of multi-source heterogeneous data update credibility according to an update lag rate of the knowledge graph. The trigger adjustment module is used for determining an extension trigger threshold of the knowledge graph according to the fitting degree of the analysis result and an actual scene. The application improves the stability of industry chain analysis.

Description

Technical Field

[0001] This invention relates to the field of knowledge graph technology, and in particular to a cross-domain data fusion and dynamic knowledge graph construction system for industrial chain analysis. Background Technology

[0002] Existing technologies can access multiple types of data, transform unstructured data into structured data through natural language processing, and achieve semantic alignment and knowledge fusion based on domain ontology design to construct basic industry chain knowledge graphs. However, existing technologies still have significant shortcomings. They lack comprehensive data coverage dimensions and targeted processing for asynchronous updates of multi-source data and semantic heterogeneity. The dynamic update mechanism is imperfect and lacks time-series storage. They also lack cross-source synchronization deviation quantification assessment and suppression mechanisms, and lack a closed-loop quality assurance system. Their scenario adaptability is limited, and they cannot adapt to the dynamic evolution of the industry chain and the complex heterogeneous characteristics of cross-domain data. This makes it easy for the graph to deviate from industry reality, making it difficult to guarantee the long-term stability, accuracy, and depth adaptability of industry chain analysis results. The stability of industry chain analysis is difficult to meet.

[0003] Chinese Patent Publication No. CN114491068A discloses a method and system for constructing an industrial park knowledge graph by integrating multi-source heterogeneous data. The method includes: collecting knowledge about the digital economy and data about digital economy industrial parks from different sources; integrating the knowledge and data into structured knowledge centered on the digital economy industrial park through data cleaning; labeling entity relationships in the structured knowledge according to a labeling strategy; constructing and training a digital economy industrial park knowledge fusion model based on a convolutional neural network; initially extracting entity relationships from the structured knowledge through the digital economy industrial park knowledge fusion model; retrieving duplicate entity relationship triples and extracting entity relationships, thereby constructing entity-relationship-entity triples for the digital economy industrial park knowledge graph, and performing knowledge fusion. However, this method and system suffer from several drawbacks. The lack of stable constraints and consistency verification mechanisms for data noise, entity ambiguity, relationship conflicts, and feature perturbations, coupled with insufficient robustness of the knowledge fusion model to local sensitive features and ambiguous boundary relationships, leads to random jumps and inconsistencies in entity relationship extraction results, resulting in decreased stability in the industrial chain analysis. Summary of the Invention

[0004] To address this, the present invention provides a cross-domain data fusion and dynamic knowledge graph construction system for supply chain analysis. This system overcomes the problems in existing technologies where the lack of stable constraints and consistency verification mechanisms for data noise, entity ambiguity, relationship conflicts, and feature disturbances leads to insufficient robustness of knowledge fusion models to local sensitive features and ambiguous boundary relationships, resulting in random jumps and inconsistencies in entity relationship extraction results and a decline in the stability of supply chain analysis.

[0005] To achieve the above objectives, this invention provides a cross-domain data fusion and dynamic knowledge graph construction system for supply chain analysis, comprising: The data processing module includes an acquisition unit for collecting multi-source heterogeneous data from multiple domain data sources and a preprocessing unit connected to the acquisition unit for preprocessing the multi-source heterogeneous data to obtain cross-domain aligned data. The knowledge graph construction module is connected to the data processing module and includes a construction unit for constructing a knowledge graph based on the cross-domain aligned data and an update unit connected to the construction unit for dynamically updating the knowledge graph based on the multi-source heterogeneous data to obtain an incremental knowledge graph. An analysis module, which is connected to the graph construction module, is used to analyze the industrial chain based on the incremental knowledge graph to obtain analysis results; A robustness determination module, which is connected to the data processing module, is used to determine whether the robustness of the multi-source heterogeneous data meets the requirements based on the multi-source data fusion characterization value determined by the noise ratio and the missing rate of the multi-source heterogeneous data. The update adjustment module, which is connected to the robustness determination module, is used to determine whether to adjust the decay coefficient of the update credibility of the multi-source heterogeneous data based on the update lag rate of the knowledge graph when the robustness of the multi-source heterogeneous data does not meet the requirements. The trigger adjustment module, which is connected to the update adjustment module, is used to determine the expansion trigger threshold of the knowledge graph based on the fit between the analysis results and the actual scenario.

[0006] Furthermore, the robustness determination module responds to the multi-source data fusion characterization value determined by the product of the noise ratio and the missing rate of the multi-source heterogeneous data to determine whether the robustness of the multi-source heterogeneous data meets the requirements.

[0007] Furthermore, the robustness determination module determines that the robustness of the multi-source heterogeneous data meets the requirements when the multi-source data fusion characterization value is greater than or equal to a preset characterization value. The robustness determination module determines that the robustness of the multi-source heterogeneous data does not meet the requirements when the multi-source data fusion characterization value is less than a preset characterization value.

[0008] Furthermore, in response to the condition that the robustness of the multi-source heterogeneous data does not meet the requirements, the update adjustment module determines whether the update stability of the knowledge graph meets the requirements based on the update lag rate of the knowledge graph.

[0009] Furthermore, the update adjustment module determines that the update stability of the knowledge graph meets the requirements when the update lag rate of the knowledge graph is less than or equal to a preset first lag rate. The update adjustment module determines that the update stability of the knowledge graph does not meet the requirements when the update lag rate of the knowledge graph is greater than a preset first lag rate.

[0010] Furthermore, the update adjustment module responds to the knowledge graph update lag rate being greater than a preset first lag rate and less than or equal to a preset second lag rate by increasing the attenuation coefficient of the reliability of multi-source heterogeneous data updates; The update adjustment module responds when the update lag rate of the knowledge graph is greater than the preset second lag rate, initially determines that the real-time update of the knowledge graph does not meet the requirements, and determines whether the real-time update of the knowledge graph meets the requirements based on the fitting degree between the analysis results and the actual scenario.

[0011] Furthermore, the increase in the decay coefficient of the reliability of the multi-source heterogeneous data update is determined by the difference between the update lag rate of the knowledge graph and the preset first lag rate.

[0012] Furthermore, in response to the condition that the update lag rate of the knowledge graph is greater than the preset second lag rate, the trigger adjustment module determines whether the real-time update of the knowledge graph meets the requirements based on the fitting degree between the analysis results and the actual scenario.

[0013] Furthermore, the trigger adjustment module determines that the real-time update of the knowledge graph meets the requirements when the fitting degree between the analysis result and the actual scene is greater than or equal to the preset fitting degree. The trigger adjustment module responds when the fit between the analysis result and the actual scenario is less than the preset fit, determines that the real-time update of the knowledge graph does not meet the requirements, and reduces the extension trigger threshold of the knowledge graph.

[0014] Furthermore, the reduction in the knowledge graph expansion trigger threshold is determined by the difference between the fitting degree of the analysis results and the actual scenario and the preset fitting degree.

[0015] Compared with existing technologies, the beneficial effects of this invention are that the system of this invention, by setting up a data processing module, a map construction module, an analysis module, a robustness determination module, an update and adjustment module, and a trigger adjustment module, determines the robustness of multi-source heterogeneous data based on the multi-source data fusion characterization value determined by the noise ratio and the missing rate of multi-source heterogeneous data. Since industry chain analysis relies on data from multiple heterogeneous sources, the data exhibits significant differences, contains a large amount of noise, and the noise ratio fluctuates dynamically with the data source scenario. The system cannot effectively suppress interference when facing noisy inputs. Through this determination... Enhancing the robustness of multi-source heterogeneous data can automatically filter out sporadic errors and identify weak signals of structural risks hidden in low-quality multi-source data. This avoids graph jitter and wasted computational resources caused by noise, making the analysis conclusions more resistant to interference and interpretable in dynamic and complex environments. The attenuation coefficient of the update credibility of multi-source heterogeneous data is adjusted according to the update lag rate of the knowledge graph. Due to the structural and semantic heterogeneity of multi-source cross-domain data and the real-time differences in the update frequency of each data source, entities, attributes, and relationships in the knowledge graph cannot be synchronously updated and consistently calibrated, leading to problems in the industry chain. Fluctuations in reasoning and analysis results can be mitigated by increasing the decay coefficient of the reliability of updates from multi-source heterogeneous data. This reduces the weight of unreliable data caused by asynchronous updates and data heterogeneity, weakening its interference with knowledge graph consistency calibration and supply chain reasoning. This suppresses fluctuations in analysis results and improves the stability and reliability of supply chain reasoning and decision-making. The knowledge graph's expansion trigger threshold can be adjusted based on the fit between the analysis results and the actual scenario. However, due to the lack of a mechanism for co-evolution with the real supply chain, it is impossible to promptly perceive industrial technological changes and the emergence of new business models through cutting-edge signals. Limited by a static and fixed ontology structure, it is difficult to reasonably model new entities and relationships. The evolution process lacks quality assurance, and the knowledge graph gradually deviates from industrial reality, leading to systematic biases and loss of long-term stability in the analysis results. Reducing the knowledge graph's expansion trigger threshold lowers the threshold for incorporating new knowledge, enabling rapid responses to cutting-edge signals such as industrial technological changes and emerging business models. Dynamically expanding the ontology structure to adapt to the modeling needs of new entities and relationships reduces the deviation between the knowledge graph and industrial reality, suppresses systematic biases in analysis results, restores and maintains the long-term stability of supply chain analysis, and improves the overall stability of supply chain analysis.

[0016] Furthermore, the system described in this invention determines the robustness of multi-source heterogeneous data by setting preset characterization values. Since supply chain analysis relies on data from multiple heterogeneous sources, the data exhibits significant differences, contains a large amount of noise, and the noise ratio fluctuates dynamically with the data source scenario. The system cannot effectively suppress interference when faced with noisy input. By determining the robustness of multi-source heterogeneous data, it can automatically filter out occasional errors and identify weak structural risk signals hidden in multi-source low-quality data, avoiding graph jitter and wasted computing resources caused by noise. This makes the analysis conclusions more resistant to interference and more interpretable in dynamic and complex environments, further improving the stability of supply chain analysis.

[0017] Furthermore, the system of the present invention adjusts the decay coefficient of the reliability of multi-source heterogeneous data updates by setting a preset first lag rate and a preset second lag rate. Due to the heterogeneity of multi-source cross-domain data in terms of structure and semantics, as well as the real-time differences in the update frequency of each data source, entities, attributes, and relationships in the knowledge graph cannot be updated synchronously and calibrated in a consistent manner, causing fluctuations in the results of supply chain reasoning and analysis. By increasing the decay coefficient of the reliability of multi-source heterogeneous data updates, the weight of untrusted data caused by asynchronous updates and data heterogeneity differences can be reduced, weakening its interference with the consistency calibration of the knowledge graph and supply chain reasoning, thereby suppressing fluctuations in analysis results, improving the stability and reliability of supply chain reasoning and decision-making, and further improving the stability of supply chain analysis.

[0018] Furthermore, the system described in this invention adjusts the expansion trigger threshold of the knowledge graph by setting a preset fitting degree. Due to the lack of a mechanism for co-evolution with the real industrial chain, it is unable to perceive industrial technology changes and the emergence of new business models in a timely manner through cutting-edge signals. Limited by the static and solidified ontology structure, it is difficult to reasonably model new entities and new relationships. The evolution process lacks quality assurance, and the knowledge graph gradually deviates from industrial reality. The systematic bias and long-term stability of the analysis results are lost. By reducing the expansion trigger threshold of the knowledge graph, the trigger threshold for the inclusion of new knowledge can be lowered, enabling a rapid response to cutting-edge signals such as industrial technology changes and emerging business models. The ontology structure can be dynamically expanded to adapt to the modeling needs of new entities and new relationships, reducing the deviation between the knowledge graph and industrial reality, suppressing the systematic bias of the analysis results, restoring and maintaining the long-term stability of industrial chain analysis, and further improving the stability of industrial chain analysis. Attached Figure Description

[0019] Figure 1 This is an overall structural diagram of the cross-domain data fusion and dynamic knowledge graph construction system for supply chain analysis, as described in an embodiment of the present invention. Figure 2This is a flowchart illustrating the process of determining the robustness of multi-source heterogeneous data in the cross-domain data fusion and dynamic knowledge graph construction system for supply chain analysis, as described in an embodiment of the present invention. Figure 3 This is a flowchart illustrating the process of determining the attenuation coefficient of the reliability of multi-source heterogeneous data updates in the cross-domain data fusion and dynamic knowledge graph construction system for supply chain analysis, as described in this embodiment of the invention. Figure 4 This is a flowchart illustrating the process of determining the real-time update of a knowledge graph in a cross-domain data fusion and dynamic knowledge graph construction system for supply chain analysis, as described in an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0021] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0022] Please see Figure 1 As shown, it is an overall structural block diagram of the cross-domain data fusion and dynamic knowledge graph construction system for industrial chain analysis according to an embodiment of the present invention.

[0023] An embodiment of the present invention provides a cross-domain data fusion and dynamic knowledge graph construction system for supply chain analysis, comprising: The data processing module includes an acquisition unit for collecting multi-source heterogeneous data from multiple domain data sources and a preprocessing unit connected to the acquisition unit for preprocessing the multi-source heterogeneous data to obtain cross-domain aligned data. The knowledge graph construction module is connected to the data processing module and includes a construction unit for constructing a knowledge graph based on the cross-domain aligned data and an update unit connected to the construction unit for dynamically updating the knowledge graph based on the multi-source heterogeneous data to obtain an incremental knowledge graph. An analysis module, which is connected to the graph construction module, is used to analyze the industrial chain based on the incremental knowledge graph to obtain analysis results; A robustness determination module, which is connected to the data processing module, is used to determine whether the robustness of the multi-source heterogeneous data meets the requirements based on the multi-source data fusion characterization value determined by the noise ratio and the missing rate of the multi-source heterogeneous data. The update adjustment module, which is connected to the robustness determination module, is used to determine whether to adjust the decay coefficient of the update credibility of the multi-source heterogeneous data based on the update lag rate of the knowledge graph when the robustness of the multi-source heterogeneous data does not meet the requirements. The trigger adjustment module, which is connected to the update adjustment module, is used to determine the expansion trigger threshold of the knowledge graph based on the fit between the analysis results and the actual scenario.

[0024] Specifically, the multi-domain data sources include supply chain procurement systems, websites of various functional departments, and mainstream news media.

[0025] Specifically, multi-source heterogeneous data includes purchase orders, logistics node status, and stock price fluctuations.

[0026] Specifically, cross-domain aligned data includes merged purchase orders, standardized logistics node status, and semantically aligned stock price fluctuations.

[0027] Specifically, the process of constructing a knowledge graph based on cross-domain aligned data involves preprocessing multi-source heterogeneous data to obtain cross-domain aligned data, transforming it into structured triples composed of entity nodes, relation edges, and attributes according to defined semantic specifications, and injecting it into a graph database to construct a semantic knowledge network that can accurately map the real industrial chain structure and support path reasoning and dynamic analysis.

[0028] Specifically, the process of dynamically updating the knowledge graph based on multi-source heterogeneous data to obtain an incremental knowledge graph involves comparing newly collected multi-source heterogeneous data with the existing knowledge graph, identifying newly added, changed, or invalid entities and relationships, and expanding the graph structure while preserving the original state through incremental injection, attribute fusion, time-sensitive labeling, and conflict resolution, thereby generating an incremental knowledge graph that reflects the latest industry reality.

[0029] Specifically, an incremental knowledge graph is a knowledge graph that includes the latest changes, obtained by updating the existing knowledge graph with real-time data.

[0030] Specifically, the process of analyzing the industrial chain based on the incremental knowledge graph to obtain analysis results involves using the latest entity, relationship, and attribute data in the incremental knowledge graph to perform industrial chain correlation reasoning, trend judgment, and risk identification, and outputting stable and reliable analysis results.

[0031] Specifically, the analysis results include interrupted purchase orders, prolonged logistical disruptions, and sharp fluctuations in stock prices.

[0032] Specifically, the industrial chain includes the semiconductor industrial chain, the integrated circuit industrial chain, and the new energy vehicle industrial chain.

[0033] Specifically, the decay coefficient of the credibility of multi-source heterogeneous data updates is a quantitative parameter that characterizes the intensity of suppression applied to low-credibility data sources caused by asynchronous updates, semantic conflicts, or structural inconsistencies when updating the knowledge graph by fusing multi-source heterogeneous data.

[0034] Specifically, the knowledge graph expansion trigger threshold is the minimum value that triggers the addition of new entity types, relation types, and ontology structure expansions to the knowledge graph.

[0035] In implementation, the system of this invention, by setting up a data processing module, a map construction module, an analysis module, a robustness determination module, an update and adjustment module, and a trigger adjustment module, determines the robustness of multi-source heterogeneous data based on the multi-source data fusion characterization value determined by the noise ratio and the missing rate of the multi-source heterogeneous data. Since industry chain analysis relies on data from multiple heterogeneous sources, the data exhibits significant differences, contains a large amount of noise, and the noise ratio dynamically fluctuates with the data source scenario. The system cannot effectively suppress interference when facing noisy inputs. Therefore, the robustness of multi-source heterogeneous data is determined through this robustness assessment. It can automatically filter out sporadic errors and identify weak structural risk signals hidden in multi-source low-quality data, avoiding graph jitter and wasted computing resources caused by noise. This makes the analysis conclusions more robust and interpretable in dynamic and complex environments. The attenuation coefficient of the update credibility of multi-source heterogeneous data is adjusted according to the update lag rate of the knowledge graph. Due to the structural and semantic heterogeneity of multi-source cross-domain data and the real-time differences in the update frequency of each data source, entities, attributes, and relationships in the knowledge graph cannot be synchronously updated and consistent, leading to problems in the results of industry chain reasoning and analysis. Fluctuations can be mitigated by increasing the decay coefficient of the reliability of multi-source heterogeneous data updates. This reduces the weight of unreliable data caused by asynchronous updates and data heterogeneity, weakening its interference with knowledge graph consistency calibration and supply chain reasoning. This suppresses fluctuations in analysis results and improves the stability and reliability of supply chain reasoning and decision-making. The knowledge graph's expansion trigger threshold can be adjusted based on the fit between the analysis results and the actual scenario. Due to the lack of a mechanism for co-evolution with the real supply chain, it is impossible to promptly perceive industrial technological changes and the emergence of new business models through cutting-edge signals. Limited by a static and fixed ontology structure, it is difficult to reasonably model new entities and relationships. The evolution process lacks quality assurance, and the knowledge graph gradually deviates from industrial reality, resulting in systematic biases and loss of long-term stability in the analysis results. Reducing the knowledge graph's expansion trigger threshold lowers the threshold for incorporating new knowledge, enabling rapid responses to cutting-edge signals such as industrial technological changes and emerging business models. Dynamically expanding the ontology structure to adapt to the modeling needs of new entities and relationships reduces the deviation between the knowledge graph and industrial reality, suppresses systematic biases in analysis results, restores and maintains the long-term stability of supply chain analysis, and improves the stability of supply chain analysis.

[0036] Please continue reading. Figure 2 As shown, it is a logical flowchart of the process of determining the robustness of multi-source heterogeneous data in the cross-domain data fusion and dynamic knowledge graph construction system for industrial chain analysis according to an embodiment of the present invention.

[0037] Specifically, the robustness determination module responds to the multiplication of the noise ratio in the multi-source heterogeneous data and the missing rate of the multi-source heterogeneous data to determine the multi-source data fusion characterization value to determine whether the robustness of the multi-source heterogeneous data meets the requirements.

[0038] Specifically, the robustness determination module determines that the robustness of the multi-source heterogeneous data meets the requirements when the multi-source data fusion characterization value is greater than or equal to a preset characterization value. The robustness determination module determines that the robustness of the multi-source heterogeneous data does not meet the requirements when the multi-source data fusion characterization value is less than a preset characterization value.

[0039] Understandably, in cross-domain data fusion and dynamic knowledge graph construction systems, the core logic of using preset characterization values ​​to assess the robustness of multi-source heterogeneous data is to transform the robustness of multi-source heterogeneous data into quantifiable characterization values ​​for judgment. The preset characterization value serves as the dividing line for determining whether the robustness of multi-source heterogeneous data meets the requirements. The preset characterization value can be set according to actual working conditions. The preset characterization value aims to ensure the stability and practicality of supply chain analysis. Optionally, the preset characterization value is determined through a limited number of experiments by evaluating the effect of different characterization values ​​on supply chain analysis. The determined preset characterization value should satisfy the condition that it is neither too small nor causes excessive interference to the supply chain analysis process. For example, the preset characterization value is generally selected within the range of [70%, 90%].

[0040] Preferably, the preset characterization value is 80% in the preferred embodiment.

[0041] Specifically, the noise ratio in multi-source heterogeneous data is the ratio of the amount of multi-source heterogeneous data containing noise to the total amount of multi-source heterogeneous data.

[0042] Specifically, the missing rate of multi-source heterogeneous data is the ratio of the amount of missing data in multi-source heterogeneous data to the total amount of data in multi-source heterogeneous data.

[0043] In practice, the system described in this invention determines the robustness of multi-source heterogeneous data by setting preset characterization values. Since supply chain analysis relies on data from multiple heterogeneous sources, the data exhibits significant differences, contains a large amount of noise, and the noise ratio fluctuates dynamically with the data source scenario. The system cannot effectively suppress interference when faced with noisy input. By determining the robustness of multi-source heterogeneous data, it can automatically filter out occasional errors and identify weak structural risk signals hidden in multi-source low-quality data. This avoids graph jitter and wasted computing resources caused by noise, enabling the analysis conclusions to have stronger anti-interference capabilities and interpretability in dynamic and complex environments, further improving the stability of supply chain analysis.

[0044] Please continue reading. Figure 3 As shown, it is a logical flowchart of the process of determining the decay coefficient of the reliability of multi-source heterogeneous data updates in the cross-domain data fusion and dynamic knowledge graph construction system for industrial chain analysis according to an embodiment of the present invention.

[0045] Specifically, the update adjustment module, in response to the condition that the robustness of the multi-source heterogeneous data does not meet the requirements, determines whether the update stability of the knowledge graph meets the requirements based on the update lag rate of the knowledge graph.

[0046] Specifically, the update adjustment module determines that the update stability of the knowledge graph meets the requirements when the update lag rate of the knowledge graph is less than or equal to a preset first lag rate. The update adjustment module determines that the update stability of the knowledge graph does not meet the requirements when the update lag rate of the knowledge graph is greater than a preset first lag rate.

[0047] Specifically, the update adjustment module responds to the knowledge graph update lag rate being greater than a preset first lag rate and less than or equal to a preset second lag rate by increasing the decay coefficient of the reliability of multi-source heterogeneous data updates; The update adjustment module responds when the update lag rate of the knowledge graph is greater than the preset second lag rate, initially determines that the real-time update of the knowledge graph does not meet the requirements, and determines whether the real-time update of the knowledge graph meets the requirements based on the fitting degree between the analysis results and the actual scenario.

[0048] Understandably, in cross-domain data fusion and dynamic knowledge graph construction systems, the use of preset first and second lag rates to characterize whether the update stability of the knowledge graph meets the requirements is based on the core logic of converting the update stability of the knowledge graph into a quantifiable lag rate for judgment. The preset first lag rate serves as the dividing line for judging whether the update stability of the knowledge graph meets the requirements, while the preset second lag rate serves as the dividing line between two reasons that cause the update stability of the knowledge graph to fail to meet the requirements. The preset first and second lag rates can be set according to actual working conditions. The preset first and second lag rates aim to ensure the stability and practicality of the supply chain analysis. Optionally, the preset first and second lag rates are determined through a limited number of experiments by evaluating the effect of different lag rates on the supply chain analysis. The determined preset first and second lag rates should be neither too small nor cause excessive interference to the supply chain analysis process. For example, the preset first lag rate is generally selected in the range of [1%, 4%], and the preset second lag rate is generally selected in the range of [5%, 8%].

[0049] Preferably, the first hysteresis rate is 2% in a preferred embodiment, and the second hysteresis rate is 6% in a preferred embodiment.

[0050] Specifically, the knowledge graph update lag rate is the ratio of the number of relations in the knowledge graph that have not been synchronously updated per unit time to the total number of relations in the knowledge graph.

[0051] Specifically, the increase in the decay coefficient of the reliability of the multi-source heterogeneous data update is determined by the difference between the update lag rate of the knowledge graph and the preset first lag rate.

[0052] Specifically, when the difference between the knowledge graph update lag rate and the preset first lag rate is within 2%, the decay coefficient of the credibility of multi-source heterogeneous data updates increases to 1.5 times the original value. When the difference between the knowledge graph update lag rate and the preset first lag rate exceeds 2%, in addition to increasing to 1.5 times the original value, for every 1% exceeding 2%, the decay coefficient of the credibility of multi-source heterogeneous data updates increases by 0.2%. For example, if the difference between the knowledge graph update lag rate and the preset first lag rate is 5%, and the current decay coefficient of the credibility of multi-source heterogeneous data updates is 3%, the increased decay coefficient of the credibility of multi-source heterogeneous data updates will be 3×1.5+0.2×3=5.1%.

[0053] In implementation, the system of this invention adjusts the decay coefficient of the reliability of multi-source heterogeneous data updates by setting a preset first lag rate and a preset second lag rate. Due to the heterogeneity of multi-source cross-domain data in terms of structure and semantics, as well as the real-time differences in the update frequency of each data source, entities, attributes, and relationships in the knowledge graph cannot be updated synchronously and calibrated in a consistent manner, causing fluctuations in the results of supply chain reasoning and analysis. By increasing the decay coefficient of the reliability of multi-source heterogeneous data updates, the weight of untrusted data caused by asynchronous updates and data heterogeneity differences can be reduced, weakening its interference with the consistency calibration of the knowledge graph and supply chain reasoning, thereby suppressing fluctuations in analysis results, improving the stability and reliability of supply chain reasoning and decision-making, and further improving the stability of supply chain analysis.

[0054] Please continue reading. Figure 4 As shown, it is a logical flowchart of the process of determining the real-time update of the knowledge graph in the cross-domain data fusion and dynamic knowledge graph construction system for industrial chain analysis according to an embodiment of the present invention.

[0055] Specifically, when the update lag rate of the knowledge graph is greater than the preset second lag rate, the trigger adjustment module determines whether the real-time update of the knowledge graph meets the requirements based on the fit between the analysis results and the actual scenario.

[0056] Specifically, the trigger adjustment module determines that the real-time update of the knowledge graph meets the requirements when the fitting degree between the analysis result and the actual scene is greater than or equal to the preset fitting degree. The trigger adjustment module responds when the fit between the analysis result and the actual scenario is less than the preset fit, determines that the real-time update of the knowledge graph does not meet the requirements, and reduces the extension trigger threshold of the knowledge graph.

[0057] Understandably, in cross-domain data fusion and dynamic knowledge graph construction systems, the core logic of using a preset fit degree to characterize whether the real-time update of the knowledge graph meets the requirements is to transform the real-time update of the knowledge graph into a quantifiable fit degree for judgment. The preset fit degree serves as the dividing line for judging whether the real-time update of the knowledge graph meets the requirements. The preset fit degree can be set according to actual working conditions. The preset fit degree aims to ensure the stability and practicality of supply chain analysis. Optionally, the preset fit degree is determined through a limited number of experiments by evaluating the effect of different fit degrees on supply chain analysis. The determined preset fit degree should satisfy the condition that it is neither too small nor will it cause excessive interference to the supply chain analysis process. For example, the preset fit degree is generally selected in the range of [75%, 95%].

[0058] Preferably, the preferred embodiment of the preset fit degree is 85%.

[0059] Specifically, the fit between the analysis results and the actual scenario is a quantitative parameter that characterizes the degree of consistency between the analysis results and real industry data.

[0060] Specifically, the reduction in the knowledge graph expansion trigger threshold is determined by the difference between the fitting degree of the analysis results and the actual scenario and the preset fitting degree.

[0061] Specifically, when the difference between the fitting degree of the analysis result and the actual scene and the preset fitting degree is within 2%, the knowledge graph expansion trigger threshold is reduced to 0.9 times the original value. When the difference between the fitting degree of the analysis result and the actual scene and the preset fitting degree exceeds 2%, the knowledge graph expansion trigger threshold is reduced by 1% for every 1% exceeding the original value, in addition to being reduced to 0.9 times the original value. For example, if the difference between the fitting degree of the analysis result and the actual scene and the preset fitting degree is 4%, and the current knowledge graph expansion trigger threshold is 5%, the reduced knowledge graph expansion trigger threshold is 5×0.9-1×2=2.5%.

[0062] In implementation, the system described in this invention adjusts the expansion trigger threshold of the knowledge graph by setting a preset fitting degree. Due to the lack of a mechanism for co-evolution with the real industrial chain, it is unable to perceive industrial technology changes and the emergence of new business models in a timely manner through cutting-edge signals. Limited by the static and solidified ontology structure, it is difficult to reasonably model new entities and new relationships. The evolution process lacks quality assurance, and the knowledge graph gradually deviates from industrial reality. The systematic bias and long-term stability of the analysis results are lost. By reducing the expansion trigger threshold of the knowledge graph, the trigger threshold for the inclusion of new knowledge can be lowered, enabling a rapid response to cutting-edge signals such as industrial technology changes and emerging business models. The ontology structure can be dynamically expanded to adapt to the modeling needs of new entities and new relationships, reducing the deviation between the knowledge graph and industrial reality, suppressing the systematic bias of the analysis results, restoring and maintaining the long-term stability of industrial chain analysis, and further improving the stability of industrial chain analysis.

[0063] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A cross-domain data fusion and dynamic knowledge graph construction system for supply chain analysis, characterized in that, include: The data processing module includes a data acquisition unit for collecting multi-source heterogeneous data from multiple domain data sources and a preprocessing unit connected to the data acquisition unit for sequentially fusing, standardizing, and semantically aligning the multi-source heterogeneous data to obtain cross-domain aligned data. The knowledge graph construction module is connected to the data processing module and includes a construction unit for constructing a knowledge graph based on the cross-domain aligned data and an update unit connected to the construction unit for dynamically updating the knowledge graph based on the multi-source heterogeneous data to obtain an incremental knowledge graph. An analysis module, which is connected to the graph construction module, is used to analyze the industrial chain based on the incremental knowledge graph to obtain analysis results; A robustness determination module, which is connected to the data processing module, is used to determine whether the robustness of the multi-source heterogeneous data meets the requirements based on the multi-source data fusion characterization value determined by the noise ratio and the missing rate of the multi-source heterogeneous data. The update adjustment module, which is connected to the robustness determination module, is used to determine the decay coefficient of the update credibility of the multi-source heterogeneous data based on the update lag rate of the knowledge graph when the robustness of the multi-source heterogeneous data does not meet the requirements. The trigger adjustment module, which is connected to the update adjustment module, is used to determine the expansion trigger threshold of the knowledge graph based on the fit between the analysis results and the actual scenario.

2. The cross-domain data fusion and dynamic knowledge graph construction system for supply chain analysis according to claim 1, characterized in that, The robustness determination module responds to the multi-source data fusion characterization value determined by the product of the noise ratio and the missing rate of the multi-source heterogeneous data to determine whether the robustness of the multi-source heterogeneous data meets the requirements.

3. The cross-domain data fusion and dynamic knowledge graph construction system for supply chain analysis according to claim 2, characterized in that, The robustness determination module determines that the robustness of the multi-source heterogeneous data meets the requirements when the multi-source data fusion characterization value is greater than or equal to a preset characterization value. The robustness determination module determines that the robustness of the multi-source heterogeneous data does not meet the requirements when the multi-source data fusion characterization value is less than a preset characterization value.

4. The cross-domain data fusion and dynamic knowledge graph construction system for supply chain analysis according to claim 3, characterized in that, The update adjustment module, in response to the condition that the robustness of the multi-source heterogeneous data does not meet the requirements, determines whether the update stability of the knowledge graph meets the requirements based on the update lag rate of the knowledge graph.

5. The cross-domain data fusion and dynamic knowledge graph construction system for supply chain analysis according to claim 4, characterized in that, The update adjustment module determines that the update stability of the knowledge graph meets the requirements when the update lag rate of the knowledge graph is less than or equal to a preset first lag rate. The update adjustment module determines that the update stability of the knowledge graph does not meet the requirements when the update lag rate of the knowledge graph is greater than a preset first lag rate.

6. The cross-domain data fusion and dynamic knowledge graph construction system for supply chain analysis according to claim 5, characterized in that, The update adjustment module responds to the knowledge graph update lag rate being greater than a preset first lag rate and less than or equal to a preset second lag rate by increasing the attenuation coefficient of the reliability of multi-source heterogeneous data updates. The update adjustment module responds when the update lag rate of the knowledge graph is greater than the preset second lag rate, initially determines that the real-time update of the knowledge graph does not meet the requirements, and determines whether the real-time update of the knowledge graph meets the requirements based on the fitting degree between the analysis results and the actual scenario.

7. The cross-domain data fusion and dynamic knowledge graph construction system for supply chain analysis according to claim 6, characterized in that, The increase in the decay coefficient of the reliability of the multi-source heterogeneous data update is determined by the difference between the update lag rate of the knowledge graph and the preset first lag rate.

8. The cross-domain data fusion and dynamic knowledge graph construction system for supply chain analysis according to claim 7, characterized in that, The trigger adjustment module responds to the condition that the update lag rate of the knowledge graph is greater than the preset second lag rate, and determines whether the real-time update of the knowledge graph meets the requirements based on the fit between the analysis results and the actual scenario.

9. The cross-domain data fusion and dynamic knowledge graph construction system for supply chain analysis according to claim 8, characterized in that, The trigger adjustment module responds when the degree of fit between the analysis result and the actual scene is greater than or equal to the preset degree of fit, and determines that the real-time update of the knowledge graph meets the requirements. The trigger adjustment module responds when the fit between the analysis result and the actual scenario is less than the preset fit, determines that the real-time update of the knowledge graph does not meet the requirements, and reduces the extension trigger threshold of the knowledge graph.

10. The cross-domain data fusion and dynamic knowledge graph construction system for supply chain analysis according to claim 9, characterized in that, The reduction in the knowledge graph expansion trigger threshold is determined by the difference between the fitting degree of the analysis results and the actual scenario and the preset fitting degree.

Images