Platform knowledge graph construction method and system based on bulk commodity transaction information

Abstract

The invention relates to the technical field of graph construction, in particular to a platform knowledge graph construction method and system based on bulk commodity transaction information. The method comprises the following steps: acquiring bulk commodity transaction information data; performing commodity transaction external association analysis on the bulk commodity transaction information data to generate commodity transaction external association data; performing special entity node marking on the bulk commodity transaction information data based on the commodity transaction external association data to obtain special entity node data; performing time sequence modeling on the bulk commodity transaction information data based on the special entity node data to generate a bulk commodity transaction dynamic time sequence graph; and performing event perception on the bulk commodity transaction dynamic time sequence graph to generate an uncontrollable node event. According to the method, through dynamic time sequence modeling, event-driven analysis and transaction behavior portrait construction, the defect that dynamic updating and risk early warning are lacked in bulk commodity transaction in the prior art is overcome, and the comprehensiveness and relevance of the knowledge graph are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of graph construction, and in particular to a method and system for constructing a platform knowledge graph based on commodity trading information. Background Art

[0002] Commodity trading information usually contains information in multiple dimensions, such as price fluctuations, trading volume, market demand, and supply chain dynamics. Traditional analysis methods find it difficult to extract effective correlations and trends from it. In order to meet this challenge, knowledge graph-based technology has emerged. It can help analyze and mine complex relationships and potential values ​​in commodity trading through relationship modeling of structured data. Initially, knowledge graph technology was mainly used in search engines and semantic web fields. With the development of artificial intelligence and big data technology, the application of knowledge graphs has gradually expanded to multiple industries such as finance, medical care, and energy. Especially in the field of commodities, trading information usually exists in multiple different data sources, and the relationship between data is relatively complex. By constructing a knowledge graph, this information can be integrated into a unified graph structure to help analysts understand market dynamics and predict price trends more intuitively. However, traditional static analysis cannot cope with market volatility and unforeseen external events (such as natural disasters). At the same time, the existing technology for modeling the behavior of commodity trading is usually relatively simple, and it is difficult to fully consider various potential risks, resulting in low comprehensiveness and relevance of knowledge graphs. Summary of the invention

[0003] Based on this, it is necessary to provide a method and system for constructing a platform knowledge graph based on commodity trading information to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a method for constructing a platform knowledge graph based on commodity trading information is provided, the method comprising the following steps: Step S1: Obtaining bulk commodity trading information data; performing commodity trading external correlation analysis on the bulk commodity trading information data to generate commodity trading external correlation data; marking the bulk commodity trading information data with special entity nodes based on the commodity trading external correlation data to obtain special entity node data; Step S2: Based on the special entity node data, the commodity trading information data is modeled in time series to generate a dynamic time series graph of commodity trading; the dynamic time series graph of commodity trading is event-aware to generate uncontrollable node events; the dynamic time series graph of commodity trading is dynamically updated with graph nodes and relationships according to the uncontrollable node events to generate an event-driven commodity trading graph; Step S3: collecting event target interaction data for the event-driven commodity trading graph to obtain transaction event interaction data; constructing a transaction behavior profile through the transaction event interaction data to generate an uncontrollable transaction behavior profile; using the uncontrollable transaction behavior profile to generate an optimal transaction decision path for the event-driven commodity trading graph to obtain an optimal transaction decision path; performing transaction risk transmission analysis on the optimal transaction decision path to generate transaction risk transmission data; performing risk warning on the optimal transaction decision path based on the transaction risk transmission data to generate commodity trading risk warning data; Step S4: Integrate the commodity trading risk warning data into the event-driven commodity trading map to obtain the commodity trading risk map; perform commodity trading risk reasoning and visualization based on the commodity trading risk map to generate a commodity trading risk report.

[0005] Through external correlation analysis, the present invention can comprehensively consider the multi-dimensional information involved in commodity trading (such as economic, political, environmental factors, etc.), thereby obtaining richer background data and helping to identify key influencing factors. Marking special entity nodes helps to focus on important market participants, commodity categories or events, and improves the accuracy and effectiveness of analysis. Time series modeling can capture the dynamic characteristics of commodity trading data over time, thereby identifying market trends. Event perception enhances the system's ability to respond to sudden or uncontrollable events (such as regular changes, disasters, etc.), and dynamically updates graph nodes and relationships to ensure the real-time and accuracy of graph data, and enhances adaptability to market changes. By collecting event interaction data and building behavioral portraits, we can deeply understand the behavior patterns of traders, especially the impact of uncontrollable factors on trading behavior, thereby providing data-based decision support, generating optimal trading decision paths and conducting risk transfer analysis, which helps to identify and evaluate potential trading risks. The risk warning function can identify risks in the trading process in advance and help users avoid or reduce losses. Integrating trading risk warning data into the trading map can achieve more comprehensive risk analysis and obtain potential risk patterns through reasoning. By visualizing the transaction risk map, decision makers can intuitively understand the risk situation of the market, so as to make more scientific and timely decisions. The generated transaction risk report further provides management with detailed risk analysis and response strategies, improving the quality of decision-making. Therefore, the present invention solves the defects of the existing technology in commodity trading, such as lack of dynamic update, risk warning and accurate decision support, through dynamic time series modeling, event-driven analysis and transaction behavior portrait construction, and improves the comprehensiveness and relevance of the knowledge map.

[0006] Preferably, step S1 comprises the following steps: Step S11: Acquire commodity trading information data, wherein the commodity trading information data includes market data, logistics data, meteorological data and financial data; Step S12: preprocessing the commodity trading information data to generate standard commodity trading information data, wherein the data preprocessing includes data cleaning, data denoising, missing value filling and data standardization; Step S13: Perform commodity transaction external correlation analysis on the logistics data and meteorological data in the standard commodity transaction information data to generate commodity transaction external correlation data; Step S14: Based on the external correlation data of commodity transactions, the geographic-meteorological knowledge graph expansion technology is used to mark the special entity nodes of the logistics data and the meteorological data to obtain the special entity node data.

[0007] The present invention can improve the accuracy and consistency of data through data preprocessing (cleaning, denoising, missing value filling and standardization), thereby reducing errors in subsequent analysis. By performing external correlation analysis of commodity transactions on logistics data and meteorological data, the potential relationship between different data sources can be discovered, helping to better understand the impact of market dynamics and external factors on transactions. By marking special entity nodes based on the external correlation data of commodity transactions using geographic-meteorological knowledge graph extension technology, the specific impact of specific influencing factors (such as weather changes, logistics conditions) on commodity transactions can be identified, which will help make more accurate market forecasts and decisions. The obtained special entity node data can provide a more detailed perspective and stronger early warning capabilities for commodity trading decisions. For example, identifying the impact of bad weather on transportation in advance can help companies take countermeasures in advance and reduce potential losses. With a deeper understanding of external correlation factors, companies can respond to market changes more quickly and improve their competitiveness and adaptability.

[0008] Preferably, step S13 includes the following steps: Step S131: extracting key logistics features from the logistics data in the standard bulk commodity transaction information data, extracting the transportation distance, transportation method and delivery location of the bulk commodity transaction, and obtaining key logistics feature data; Step S132: extracting meteorological key features from the meteorological data in the standard commodity trading information data, extracting the temperature, precipitation and seasonal changes of commodity trading, and obtaining meteorological key feature data; Step S133: Calculate the Pearson correlation coefficient of the logistics key feature data and the meteorological key feature data, and construct an external correlation matrix based on the Pearson correlation coefficient to generate an external correlation matrix for bulk commodity transactions; Step S134: performing correlation strength analysis on the external correlation matrix of bulk commodity transactions to generate external correlation strength data; performing correlation feature extraction on the external correlation matrix of bulk commodity transactions based on the external correlation strength data to generate external correlation data of commodity transactions.

[0009] The present invention can systematically identify the main factors that have a direct impact on commodity trading by extracting key features (such as transportation distance, transportation mode, delivery location, temperature, precipitation and seasonal changes) from logistics and meteorological data, and provide accurate basic data for subsequent analysis. By calculating the Pearson correlation coefficient, the linear correlation between logistics characteristics and meteorological characteristics can be effectively evaluated, which will help to discover potential external correlation factors, such as the efficiency differences of certain transportation modes under specific meteorological conditions, and help enterprises make optimal choices. The generation of external correlation matrix and correlation strength data can reveal the interaction between different factors, which provides a scientific basis for identifying potential risks (such as the impact of extreme weather on transportation). Enterprises can use this data relationship to make early warnings and preventive measures to improve operational efficiency and the ability to respond to emergencies. The external correlation data of commodity transactions generated by correlation feature extraction provides a more refined perspective for market analysis and decision-making. For example, based on these external data, enterprises can flexibly adjust transportation plans and inventory management strategies in the case of seasonal climate changes or changes in transportation routes. By performing correlation strength analysis on the external correlation matrix, data support can be provided for building a more accurate prediction model.

[0010] Preferably, step S2 comprises the following steps: Step S21: Based on the special entity node data, time series modeling is performed on the standard bulk commodity trading information data to generate a dynamic time series graph of bulk commodity trading; node change rate is calculated on the dynamic time series graph of bulk commodity trading to obtain the node change rate of bulk commodity trading; Step S22: screening key nodes of the dynamic time series graph of commodity trading by the change rate of commodity trading nodes to obtain uncontrollable trading nodes; performing event perception on the dynamic time series graph of commodity trading according to the uncontrollable trading nodes to generate uncontrollable node events; Step S23: dynamically update the nodes and relationships of the dynamic time series graph of bulk commodity transactions according to the uncontrollable node events, and generate an event-driven bulk commodity transaction graph.

[0011] The dynamic time series graph of commodity transactions generated by the present invention through time series modeling can reveal the timeliness and change trend of the transaction process. The calculation of the node change rate helps to identify the dynamic changes in the transaction activities, so as to accurately capture the influence of factors such as market fluctuations, price changes, and changes in transaction frequency. By screening out the key nodes in the transaction graph and identifying the uncontrollable transaction nodes, a basis is provided for identifying the key links in the transaction system that are affected by external uncontrollable factors (such as market emergencies, regular changes, etc.), which helps to carry out risk warning and adjust the transaction strategy in time. Through the generation of uncontrollable node events, enterprises can realize real-time perception of emergencies in the dynamic time series graph of commodity transactions. This event perception capability enables enterprises to capture external factors that affect transactions in advance (such as political risks, natural disasters, etc.), so as to make rapid responses. The dynamic update of graph nodes and relationships enables the graph to be adjusted in real time according to the occurrence of actual transaction events, which not only enhances the responsiveness of the transaction system to real-time information, but also ensures that the transaction graph always maintains the latest and most accurate data under changing market conditions, supporting more flexible decision-making. The event-driven commodity trading map can quickly adjust trading strategies and resource allocation after an event occurs, thereby improving the company's response speed and adaptability to market changes. By continuously updating and optimizing the map, companies can promptly discover potential opportunities or risks, optimize trading decisions, and reduce the risk of decision-making errors.

[0012] Preferably, performing event perception on the dynamic time series graph of commodity trading based on the uncontrollable trading nodes includes: The external environment data is matched with the dynamic time series graph of commodity transactions according to the uncontrollable transaction nodes to obtain the external environment data; Trigger analysis of uncontrollable transaction nodes based on external environment data to obtain event trigger data; Perform dynamic event prediction on event triggering data to obtain dynamic event prediction data; Conduct market response analysis on dynamic event prediction data to obtain market response data; Based on the market response data, risk assessment and decision support are carried out on uncontrollable transaction nodes to obtain uncontrollable node events.

[0013] By matching external environmental data, the present invention can ensure that the analysis of uncontrollable transaction nodes not only relies on internal data, but also takes into account external factors (such as regular changes, international economic trends, natural disasters, etc.), so that event perception is more comprehensive and accurate. This combination of external environmental data can provide a richer contextual background, which helps to deeply understand the reasons for changes in transaction dynamics. Through trigger analysis based on external environmental data, potential risks or precursors of important events can be identified in time, which provides an early warning mechanism for commodity trading and helps companies to respond to external shocks and avoid risks in advance. For example, a certain change in climate conditions causes a delay in a certain transportation route, and the analysis of triggering this event can help companies prepare countermeasures in advance. Dynamic event prediction of event trigger data helps to foresee the development trend of potential events in a complex market environment, which provides decision makers with data-driven prediction capabilities and enhances the ability to respond to emergencies. Through accurate prediction, companies can be fully prepared before trading, thereby reducing risks and seizing potential market opportunities. Through market response analysis, we can understand the specific impact of external events on the market (such as price fluctuations, changes in supply and demand, etc.). This analysis enables companies to better evaluate market changes caused by external shocks, adjust strategies according to market reactions, and optimize resource allocation. For example, if regular changes in a certain region lead to a sharp increase in demand for goods in that region, companies can make corresponding supply chain adjustments based on market response data.

[0014] Preferably, step S3 comprises the following steps: Step S31: collecting event target interaction data on the event-driven commodity trading graph to obtain trading event interaction data; performing trading game analysis on the trading event interaction data to generate user trading game data; Step S32: constructing a transaction behavior profile for the transaction event interaction data according to the user transaction game data to generate an uncontrollable transaction behavior profile; using the uncontrollable transaction behavior profile to identify transaction behavior patterns of the event-driven commodity transaction graph to generate transaction behavior pattern data; Step S33: Generate an optimal trading decision path for the event-driven commodity trading graph through the trading behavior pattern data to obtain the optimal trading decision path; perform anomaly detection on the optimal trading decision path to generate optimal trading decision anomaly detection data; Step S34: Perform transaction risk transfer analysis based on the optimal transaction decision anomaly detection data to generate transaction risk transfer data; perform risk warning on the optimal transaction decision path based on the transaction risk transmission data to generate commodity transaction risk warning data.

[0015] The present invention can fully understand the strategies and motivations of different market participants (such as buyers, sellers, suppliers, etc.) in the transaction process through the collection and game analysis of transaction event interaction data, which helps to reveal the potential transaction patterns and game structures in the market and provide basic data for subsequent transaction behavior analysis. The user transaction game data obtained through transaction game analysis can establish a unique transaction behavior portrait for each participant, which enables enterprises to deeply understand the behavior patterns of different users in the market, thereby providing a basis for personalized risk management and decision-making. The construction of uncontrollable transaction behavior portraits can help identify those transaction behaviors affected by uncontrollable factors and enhance the enterprise's ability to identify potential risks. Using uncontrollable transaction behavior portraits to identify transaction patterns can reveal the mainstream transaction behaviors and their evolution trends in the market. This pattern recognition helps to improve sensitivity to market fluctuations and transaction opportunities, and improve the accuracy of market forecasts and decisions. For example, identifying a specific transaction pattern indicates that the market is about to experience price fluctuations or changes in supply and demand. The optimal trading decision path generated by trading behavior pattern data can help enterprises make more efficient and accurate trading decisions in a complex market environment. This not only reduces the error of human decision-making, but also can automatically generate a decision path with strong adaptability and resilience based on historical behavior patterns, thereby improving the operational efficiency and market response capabilities of enterprises. By performing anomaly detection on the optimal trading decision path, potential abnormal trading behaviors or market fluctuations can be discovered in a timely manner. For example, certain extreme market conditions or unusual trading behaviors indicate that the market is about to undergo major changes. Anomaly detection data provides enterprises with real-time risk monitoring and early warning, thereby helping them make timely adjustments and avoid losses. Based on the anomaly detection results of the optimal trading decision path, trading risk transmission analysis can reveal the risk transmission chain between different trading nodes. This analysis can help enterprises understand how a risk event spreads from one node to other nodes, thereby forming a chain reaction, which helps to comprehensively assess potential risks and take preventive measures in a timely manner.

[0016] Preferably, using the uncontrollable transaction behavior profile to identify the transaction behavior pattern of the event-driven commodity transaction map includes: Extract commodity transaction price characteristics from the uncontrollable transaction behavior portrait to obtain commodity transaction price characteristic data; perform commodity transaction price fluctuation analysis on the commodity transaction price characteristic data through the relative strength index to generate commodity transaction price characteristic data; The commodity transaction price characteristic data is used to analyze the transaction volume of the uncontrollable transaction behavior portrait to generate the uncontrollable transaction volume; the commodity transaction price characteristic data and the uncontrollable transaction volume are used to identify the first transaction behavior pattern of the event-driven commodity transaction map to obtain the short-term manipulation behavior identification pattern; Confirm the price fundamentals of commodity transaction price feature data to obtain commodity price fundamentals data; perform deviation analysis on commodity transaction price feature data and commodity price fundamentals data to generate price fundamentals deviation data; use sentiment index to perform emotional evaluation on price fundamentals deviation data to generate commodity transaction emotional data; According to the commodity trading emotional data, the second trading behavior pattern is identified on the event-driven commodity trading map to obtain the emotional trading behavior identification pattern; the short-term manipulation behavior identification pattern and the emotional trading behavior identification pattern are integrated to obtain the trading behavior pattern data.

[0017] The present invention can accurately identify the changing trend of market prices through commodity transaction price feature extraction, relative strength index analysis and price fluctuation assessment. This analysis helps to capture abnormal patterns of price fluctuations, especially short-term manipulation behaviors, and can provide early warnings for enterprises to avoid losses caused by excessive price fluctuations. By analyzing the uncontrollable trading volume and combining the commodity transaction price feature data, the changing pattern of trading volume under price fluctuations can be identified, which provides support for identifying potential market manipulation behaviors. For example, a certain price fluctuation accompanied by abnormal trading volume indicates the existence of manipulation risks or abnormal market behaviors. By combining the commodity transaction price feature data and the uncontrollable trading volume, potential short-term manipulation behavior patterns can be identified. This pattern recognition helps to reveal signs of market manipulation, take corresponding risk prevention measures in advance, and avoid investment losses or market instability caused by price manipulation. By confirming the fundamentals of the commodity transaction price feature data and combining the price deviation analysis, the reasons behind the price fluctuations can be more comprehensively understood. If the price fluctuation deviates from the support of the fundamentals, it indicates that there are abnormal fluctuations in the market, which provides a more powerful basis for price risk assessment for enterprises. The introduction of the sentiment index can quantitatively analyze the emotional factors in price fluctuations. The impact of market sentiment on price fluctuations cannot be ignored. Emotional trading pattern recognition helps capture price fluctuations driven by investor sentiment and helps companies identify the stages of overheated or overcooled market sentiment, thereby making more rational trading decisions. By integrating short-term manipulation behavior recognition patterns and emotional trading behavior recognition patterns, comprehensive pattern recognition of trading behaviors can be performed. This comprehensive recognition helps to grasp market dynamics from multiple dimensions, not only capturing manipulation behaviors, but also identifying the impact of emotional fluctuations, providing companies with more accurate market trend judgments.

[0018] Preferably, the transaction risk transfer analysis based on the optimal transaction decision anomaly detection data includes: Based on the abnormal detection data of the optimal trading decision, the trading risk transmission index is designed to obtain the key indicators of trading risk transmission, where the key indicators of trading risk transmission include risk exposure, transmission rate and risk concentration; the optimal trading decision abnormal detection data is matched with the indicator similarity using risk exposure, transmission rate and risk concentration to obtain the trading risk indicator matching data; Mapping the transaction risk indicator matching data to the event-driven commodity trading graph to construct the associated risk indicator path and generate transaction risk path data; marking the associated nodes of the transaction risk path data to obtain nodes with the same risk indicator; The propagation distance of nodes with the same risk index is calculated to obtain the node risk propagation distance data; the transaction risk transmission analysis is performed on the optimal transaction decision anomaly detection data through the node risk propagation distance data to generate transaction risk transmission data.

[0019] The present invention can evaluate the risk exposure of each node and transaction behavior in the market through the risk exposure design in the key indicator of transaction risk transmission, which helps enterprises accurately identify areas or transaction behaviors with high risk exposure, so as to take appropriate countermeasures to reduce potential losses. Through the design of the transmission rate, the transmission speed of transaction risks between different nodes can be monitored in real time. The design provides a dynamic risk warning function for the trading system, which helps to timely discover the diffusion trend of risks. By tracking the transmission rate of risks, enterprises can effectively control market risks and avoid major losses caused by rapid diffusion. The risk concentration index can reveal potential risk concentration areas in the market, which is crucial for effective monitoring of high-risk markets. By identifying which nodes or transaction behaviors have high risk aggregation, enterprises can adopt strategies such as diversified investment and hedging in advance to reduce systemic risks. By matching the optimal transaction decision anomaly detection data with risk exposure, transmission rate and aggregation, enterprises can perform multi-dimensional risk assessment, which not only improves the ability to identify potential risks in a complex market environment, but also ensures that transaction decisions are based on comprehensive data analysis, avoiding the deviation caused by single indicator analysis. Mapping risk indicators to event-driven commodity trading maps for path construction can provide companies with a clear risk propagation path map, which enables companies to have an in-depth understanding of the risk transmission process from the source to the final node, thereby formulating more accurate risk management strategies.

[0020] Preferably, step S4 comprises the following steps: Step S41: Integrate the commodity trading risk warning data into the event-driven commodity trading map to obtain the commodity trading risk map; set a query window for the commodity trading risk map to generate a commodity trading query window; Step S42: The commodity trading risk map is used to infer and visualize the commodity trading risk through the commodity trading query window, thereby generating a commodity trading risk report.

[0021] The present invention integrates the commodity trading risk warning data into the event-driven commodity trading map, and can concentrate the scattered risk information in one map for analysis and monitoring. This integrated processing method enables enterprises to fully understand all risk factors, avoid missing or repeated analysis, and improve risk management efficiency. By setting a commodity trading query window, users can dynamically choose to view specific risk data according to their needs, which is crucial for coping with the ever-changing market environment. The flexibility of the query window allows enterprises to obtain the required risk information in real time, respond quickly, and avoid missing important risk warning signals. Through the commodity trading query window, risk reasoning can be performed, and enterprises can make accurate risk predictions and analyses based on historical data, real-time data, and event-driven analysis in the map. This reasoning ability allows enterprises to gain insight into the impact of potential market fluctuations, price fluctuations, or regular changes, and enhance forward-looking decision-making capabilities. The generated commodity trading risk report can intuitively display various risk data, reasoning processes, and their results, helping decision makers quickly understand the source, propagation path, and impact of risks. This visualization method reduces the complexity of information transmission, increases the readability of the report, and enhances the decision makers' understanding and response capabilities to complex data.

[0022] In this specification, a platform knowledge graph construction system based on commodity trading information is provided, which is used to execute the above-mentioned platform knowledge graph construction method based on commodity trading information. The platform knowledge graph construction system based on commodity trading information includes: The node setting module is used to obtain commodity trading information data; perform commodity trading external correlation analysis on the commodity trading information data to generate commodity trading external correlation data; perform special entity node marking on the commodity trading information data based on the commodity trading external correlation data to obtain special entity node data; The graph construction module is used to: perform time series modeling on commodity trading information data based on special entity node data to generate a dynamic time series graph of commodity trading; perform event perception on the dynamic time series graph of commodity trading to generate uncontrollable node events; dynamically update the graph nodes and relationships of the dynamic time series graph of commodity trading according to the uncontrollable node events to generate an event-driven commodity trading graph; The risk analysis module is used to collect event target interaction data for the event-driven commodity trading map to obtain transaction event interaction data; construct transaction behavior profiles through transaction event interaction data to generate uncontrollable transaction behavior profiles; use uncontrollable transaction behavior profiles to generate the optimal transaction decision path for the event-driven commodity trading map to obtain the optimal transaction decision path; perform transaction risk transmission analysis on the optimal transaction decision path to generate transaction risk transmission data; perform risk warning on the optimal transaction decision path based on the transaction risk transmission data to generate commodity trading risk warning data; The graph reasoning module is used to integrate commodity trading risk warning data into the event-driven commodity trading graph to obtain a commodity trading risk graph; perform commodity trading risk reasoning and visualization based on the commodity trading risk graph to generate a commodity trading risk report.

[0023] The beneficial effect of the present invention is that the node setting module provides detailed and accurate external correlation data for subsequent risk identification and graph construction by obtaining commodity trading information and conducting external correlation analysis of commodity transactions. This in-depth external correlation analysis can help identify external factors that affect transactions, such as market changes, logistics, weather, etc., thereby enhancing the accuracy of decision-making. The time series modeling function of the graph construction module can dynamically reflect the changes of various events in the commodity trading process and adjust the graph nodes and relationships in real time, which enables the system to perceive and update uncontrollable node events in real time during the transaction process, establish an event-driven graph model, and support dynamic decision-making. The risk analysis module can conduct in-depth analysis of the potential risks of transaction decisions by constructing transaction behavior portraits and analyzing the optimal transaction decision path, provide the most optimized transaction path selection, and conduct systematic early warning through transaction risk transmission analysis. The core of this module is to quantitatively analyze the transmission process of transaction risks, thereby providing early warning of emerging risks. The graph reasoning module integrates an event-driven commodity trading risk graph and provides a graph reasoning function, which supports the visualization of commodity trading risks through the reasoning process. This module not only provides deep data mining capabilities, but also can pass complex risk analysis processes to decision makers through visual reports, thereby improving the transparency and accuracy of decision-making. Therefore, the present invention solves the defects of the existing technology in commodity trading, which lacks dynamic updates, risk warnings, and accurate decision-making support, through dynamic time series modeling, event-driven analysis, and transaction behavior portrait construction, and improves the comprehensiveness and relevance of the knowledge graph. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A schematic diagram of the steps of a method for constructing a platform knowledge graph based on commodity trading information; Figure 2 for Figure 1 Detailed implementation steps of step S2 in the flowchart; Figure 3 for Figure 1 Detailed implementation steps of step S3 in FIG. The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0025] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0026] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0027] It should be understood that although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit, and the term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0028] To achieve this, please refer to Figures 1 to 3 , a method for constructing a platform knowledge graph based on commodity trading information, the method comprising the following steps: Step S1: Obtaining bulk commodity trading information data; performing commodity trading external correlation analysis on the bulk commodity trading information data to generate commodity trading external correlation data; marking the bulk commodity trading information data with special entity nodes based on the commodity trading external correlation data to obtain special entity node data; Step S2: Based on the special entity node data, the commodity trading information data is modeled in time series to generate a dynamic time series graph of commodity trading; the dynamic time series graph of commodity trading is event-aware to generate uncontrollable node events; the dynamic time series graph of commodity trading is dynamically updated with graph nodes and relationships according to the uncontrollable node events to generate an event-driven commodity trading graph; Step S3: collecting event target interaction data for the event-driven commodity trading graph to obtain transaction event interaction data; constructing a transaction behavior profile through the transaction event interaction data to generate an uncontrollable transaction behavior profile; using the uncontrollable transaction behavior profile to generate an optimal transaction decision path for the event-driven commodity trading graph to obtain an optimal transaction decision path; performing transaction risk transmission analysis on the optimal transaction decision path to generate transaction risk transmission data; performing risk warning on the optimal transaction decision path based on the transaction risk transmission data to generate commodity trading risk warning data; Step S4: Integrate the commodity trading risk warning data into the event-driven commodity trading map to obtain the commodity trading risk map; perform commodity trading risk reasoning and visualization based on the commodity trading risk map to generate a commodity trading risk report.

[0029] Through external correlation analysis, the present invention can comprehensively consider the multi-dimensional information involved in commodity trading (such as economic, political, environmental factors, etc.), thereby obtaining richer background data and helping to identify key influencing factors. Marking special entity nodes helps to focus on important market participants, commodity categories or events, and improves the accuracy and effectiveness of analysis. Time series modeling can capture the dynamic characteristics of commodity trading data over time, thereby identifying market trends. Event perception enhances the system's ability to respond to sudden or uncontrollable events (such as regular changes, disasters, etc.), and dynamically updates graph nodes and relationships to ensure the real-time and accuracy of graph data, and enhances adaptability to market changes. By collecting event interaction data and building behavioral portraits, we can deeply understand the behavior patterns of traders, especially the impact of uncontrollable factors on trading behavior, thereby providing data-based decision support, generating optimal trading decision paths and conducting risk transfer analysis, which helps to identify and evaluate potential trading risks. The risk warning function can identify risks in the trading process in advance and help users avoid or reduce losses. Integrating trading risk warning data into the trading map can achieve more comprehensive risk analysis and obtain potential risk patterns through reasoning. By visualizing the transaction risk map, decision makers can intuitively understand the risk situation of the market, so as to make more scientific and timely decisions. The generated transaction risk report further provides management with detailed risk analysis and response strategies, improving the quality of decision-making. Therefore, the present invention solves the defects of the existing technology in commodity trading, such as lack of dynamic update, risk warning and accurate decision support, through dynamic time series modeling, event-driven analysis and transaction behavior portrait construction, and improves the comprehensiveness and relevance of the knowledge map.

[0030] In the embodiment of the present invention, reference Figure 1 As shown, it is a schematic diagram of the steps of a method for constructing a platform knowledge graph based on bulk commodity trading information of the present invention. In this example, the method for constructing a platform knowledge graph based on bulk commodity trading information includes the following steps: Step S1: Obtaining bulk commodity trading information data; performing commodity trading external correlation analysis on the bulk commodity trading information data to generate commodity trading external correlation data; marking the bulk commodity trading information data with special entity nodes based on the commodity trading external correlation data to obtain special entity node data; In an embodiment of the present invention, step S1: obtaining commodity trading information data; performing commodity trading external correlation analysis on the commodity trading information data to generate commodity trading external correlation data; marking the commodity trading information data with special entity nodes based on the commodity trading external correlation data to obtain special entity node data; and obtaining the data from multiple channels such as trading platforms, industry reports, market research institutions, and statistical data through API interfaces. Through data mining and association rule analysis (such as Apriori algorithm, FP-growth algorithm, etc.), external factors related to commodity trading are identified, and these external factors include market news, regular changes, weather conditions, transportation conditions, geopolitics, etc. Clean and standardize the original data, remove noise data, and ensure the accuracy and consistency of the data. Use machine learning algorithms (such as cluster analysis, graph mining, and association rule analysis) to discover potential correlation patterns in commodity trading data. Based on the results of the external correlation analysis, a new data set is generated, which includes the correlation information between the trading data and the external factors. For example, the correlation between the trading volume and a specific news event, weather conditions, or market sentiment. These external correlation data are stored as structured data (such as a relational database or a NoSQL database) to facilitate subsequent processing and analysis. Through external related data, entity nodes with special significance in commodity trading data are identified. These nodes represent important market participants, key trading events, special commodities or market activities. Add tags to these nodes, such as using graph structures to represent the relationship between different entities, and use node attributes or labels to represent their particularity in transactions. These special entity nodes can be visualized and analyzed using graph databases (such as Neo4j) or network analysis tools (such as Gephi). Combined with machine learning models, nodes can be classified or clustered to gain a deep understanding of transactions. The marked special entity node information is saved in the final dataset and provides basic data for further analysis. These special entity nodes can be combined with other data sources (such as macroeconomic indicators, company financial report data, etc.) for more in-depth trend analysis, risk assessment or predictive modeling.

[0031] Step S2: Based on the special entity node data, the commodity trading information data is modeled in time series to generate a dynamic time series graph of commodity trading; the dynamic time series graph of commodity trading is event-aware to generate uncontrollable node events; the dynamic time series graph of commodity trading is dynamically updated with graph nodes and relationships according to the uncontrollable node events to generate an event-driven commodity trading graph; In an embodiment of the present invention, a time series data set is constructed by sorting commodity trading information according to timestamps. Each data point contains time, transaction volume, price, transaction party, commodity type, external related information, etc. Statistical methods (such as moving average, weighted sliding average) or machine learning methods (such as LSTM, GRU, etc.) are used to extract time series features from the data, and these features will help capture the dynamic trend of the commodity market. A time series modeling method based on a graph neural network (GNN) can be used to regard transaction information as nodes in a graph, and model the relationship changes between nodes through the time series evolution of the graph. Specifically, a temporal graph neural network (Temporal GCN or TGN) can be used to capture the time-varying characteristics of nodes and relationships in the commodity trading process. Each special entity node (such as a trading market, commodity, important participant, etc.) is used as a node in the graph, and the relationship between nodes (such as transactions, collaboration, price changes, etc.) is used as an edge, and a dynamic graph is established through time series data. In the dynamic graph, the attributes of nodes and edges change over time. For example, the attributes of a node may include transaction volume, price, etc., and the attributes of an edge may represent the type of transaction, the strength of a relationship, etc. Each node and edge is updated over time based on market changes. Use graph databases (such as Neo4j) or graph computing frameworks (such as PyTorch Geometric, DGL) to implement the construction and evolution of dynamic graphs. Design event detection mechanisms using the time-varying characteristics of time-series graphs. For example, abnormal fluctuations in commodity trading can be identified through mutation point detection algorithms (such as CUSUM, LSTM-based anomaly detection), or key events can be marked using external correlation data (such as macroeconomic laws, emergencies, etc.). According to the characteristics of the commodity market, deep learning or traditional machine learning methods (such as anomaly detection, time series analysis) can be used for event perception. For example, graph-based convolutional networks (GCN) or time-series graph neural networks (TGN) are used to perceive events in time-series graphs. Uncontrollable event nodes are identified from dynamic time-series graphs, which usually have a significant impact on trading dynamics, such as sudden market fluctuations, regular changes, and supply chain disruptions. Through the results of event perception, uncontrollable events driven by external factors in the dynamic time series graph of commodity trading are identified. These events usually involve sudden market changes, such as large price fluctuations, regular announcements, natural disasters, etc. Uncontrollable node events are labeled and key information such as event type, occurrence time, and impact range are recorded. This information will provide a basis for subsequent dynamic updates of the graph. According to uncontrollable node events, nodes and edges in the graph are dynamically updated. For example: add event-related nodes and modify the properties of existing nodes (such as price, trading volume, etc.). Update the relationship between nodes after the event occurs (such as adding or deleting edges, adjusting edge weights, etc.).For example, a certain rule leads to a surge in the trading volume of a specific commodity. At this time, the relationship strength of the relevant trading nodes and the weight of the edges can be increased accordingly. Graph-based algorithms (such as graph convolution, graph attention mechanism) or incremental graph processing methods are used to achieve dynamic updates of the graph. Based on the dynamic update of uncontrollable node events, a complete commodity trading graph containing the latest trading dynamics and event-driven changes is generated. This graph not only reflects the long-term trading trends of the market, but also promptly reflects market fluctuations caused by emergencies. The final generated event-driven commodity trading graph is visualized to help analysts quickly identify market changes and trends. Visualization tools such as Gephi, NetworkX, D3.js, etc. can be used to show the dynamic changes of nodes and edges in the graph. Based on the event-driven commodity trading graph, subsequent risk assessment and market trend prediction are carried out. By analyzing the event-driven node and relationship changes, potential market risks can be discovered and corresponding response strategies can be formulated. The dynamic graph is applied to the commodity trading decision-making process to help market participants (such as traders, investors, enterprises, etc.) make real-time decisions and adjust strategies.

[0032] Step S3: collecting event target interaction data for the event-driven commodity trading graph to obtain transaction event interaction data; constructing a transaction behavior profile through the transaction event interaction data to generate an uncontrollable transaction behavior profile; using the uncontrollable transaction behavior profile to generate an optimal transaction decision path for the event-driven commodity trading graph to obtain an optimal transaction decision path; performing transaction risk transmission analysis on the optimal transaction decision path to generate transaction risk transmission data; performing risk warning on the optimal transaction decision path based on the transaction risk transmission data to generate commodity trading risk warning data; In the embodiment of the present invention, "event target interaction data" is defined as the interactive information between different market participants or trading nodes after a specific event occurs. For example, after a certain law changes, the market transaction reaction caused by the fluctuation of commodity prices. Use API or data interface to obtain market data related to the event in real time (such as trading volume, price fluctuations, buy and sell orders, news reports, etc.). Collect data from external sources (such as news, social media, macroeconomic laws, etc.), especially those uncontrollable events that affect the commodity market. Integrate these interaction data into the event-driven commodity trading graph to capture the interaction pattern between participants after the event occurs. According to the interaction between nodes (market participants) and edges (transaction behaviors) in the graph, transaction event interaction data is generated. These data usually include the trading behavior of participants (such as buying and selling decisions, price response, etc.), transaction volume, timestamps, and response methods of external events. Store these interaction data as time series data or event stream data for subsequent analysis. Distributed databases or stream processing platforms (such as Apache Kafka, Flink) can be used to efficiently process and store real-time interaction data. Based on the collected transaction event interaction data, the behavior of each market participant is modeled through machine learning and graph analysis methods. The transaction behavior profile should include the trading mode, decision-making style, risk preference, reaction speed and other characteristics of the participants. Extract key features that affect trading behavior, such as trading frequency, bid-ask spread, price fluctuation response, etc. Use cluster analysis (such as K-means, DBSCAN) or deep learning models (such as Autoencoder, LSTM) to classify and model trading behaviors. Pay special attention to trading behaviors caused by external uncontrollable events (such as regular changes, natural disasters, market crises, etc.), which are usually highly volatile and sudden, so they need to be modeled separately. Use anomaly detection algorithms (such as Isolation Forest, One-Class SVM) to identify and mark abnormal trading behaviors caused by uncontrollable events and generate uncontrollable trading behavior profiles. Use uncontrollable trading behavior profiles to generate the optimal trading decision path in the event-driven commodity trading map. The optimal decision path should take into account the current market status, the impact of external events, the behavioral characteristics of participants, and risk management. Define the criteria for the optimal path, such as profit maximization, risk minimization, and optimal trading timing. Reinforcement learning algorithms (such as Q-learning, Deep Q-Networks) or shortest path algorithms (such as Dijkstra algorithm) can be used to generate the optimal trading decision path. The optimal path should be adjusted at any time according to the dynamic changes in the graph and the occurrence of uncontrollable events. Through the path generation algorithm, one or more optimal trading decision paths are obtained. These paths will provide market participants with the best trading strategies driven by specific events, including the best entry point, exit point, capital allocation, risk control, etc.Compare the optimal decision path with historical data, evaluate the effectiveness of the path, and further optimize the decision path based on historical backtesting results. Consider the interaction and risk transmission process between various market participants. For example, a surge in the price of a certain commodity triggers a large number of follow-up transactions, thereby exacerbating market volatility. Construct a risk transmission link model to analyze how the risk of one node is transmitted to other nodes. Propagation models (such as the SIR model and the risk propagation model in complex networks) can be used to analyze the propagation path of transaction risks. Quantify each risk transmission link and evaluate its contribution to the overall market risk. For example, use risk measurement methods such as Value at Risk (VaR) or Conditional Value at Risk (CVaR) to quantify risks. Based on the results of transaction risk transmission analysis, generate transaction risk transmission data, which includes information such as risk impact between different transaction nodes (participants), risk transmission path, and risk exposure of each node. Store risk transmission data as graph structure or matrix data for further analysis and visualization. Risk data can be visualized using graph analysis tools (such as Gephi, NetworkX) or BI tools (such as Tableau, Power BI).

[0033] Step S4: Integrate the commodity trading risk warning data into the event-driven commodity trading map to obtain the commodity trading risk map; perform commodity trading risk reasoning and visualization based on the commodity trading risk map to generate a commodity trading risk report.

[0034] In an embodiment of the present invention, risk warning data related to commodity trading are collected, including market volatility, abnormal price fluctuations, regular changes, supply chain disruptions, global economic events, etc., which are usually derived from market monitoring systems, external data providers, social media analysis, etc. The risk warning data is standardized and converted into a form suitable for the graph structure. The data includes information such as risk type (such as market risk, regular risk, supply chain risk, etc.), risk intensity, occurrence time, warning source, etc. The risk warning data is integrated into the event-driven commodity trading graph. The risk warning data can be stored as node attributes or edge attributes in the graph. For example: the node attribute adds risk exposure, potential loss and other indicators to the participants in the commodity market (such as traders, suppliers, etc.). The edge attribute adds risk-related attributes to the connection between different transactions, such as transaction frequency, potential fluctuations, transaction risks, etc. The risk warning data is integrated and the graph is expanded using a graph database (such as Neo4j, ArangoDB) or a distributed graph computing framework (such as Apache Flink, Pregel). Based on the risk warning data integrated in step S41, a commodity trading risk graph is generated. In this graph, nodes represent entities such as trading markets, trading participants, and commodities, while edges represent trading relationships, market impacts, and so on. Risk attributes are added to each node, such as risk exposure, potential losses, and impact range. Risk indicators are added to trading relationships, such as trading volatility and risk transmission paths. The commodity trading risk graph needs to be updated in real time. As market dynamics and external events change, new risk data should be continuously integrated into the graph. Real-time updates and dynamic evolution of the graph are achieved through stream processing frameworks (such as Apache Kafka and Flink). Graph reasoning methods (such as graph convolutional neural networks (GCNs) and graph attention networks (GATs)) are used to reason about trading risks. These reasoning methods can infer potential risk transmission paths and risk accumulation areas based on the attributes of nodes and edges in the graph. A graph-based risk propagation model is used to deduce how risks are propagated to other nodes through the edges in the graph when risks occur at a certain node (such as a trading participant or market). For example, the SIR model or the propagation model in complex networks is used to simulate the spread of risks. Apply graph algorithms (such as breadth-first search (BFS), depth-first search (DFS)) or graph neural networks (GNN) to perform risk reasoning and identify potential risk transmission links and the most exposed market areas. Use graph analysis tools (such as Neo4j's graph analysis library and PyTorch Geometric) for graph reasoning and risk speculation. Visualize the commodity trading risk map to help analysts more intuitively understand the risk relationship between various market participants and transactions. Use graph visualization tools such as Gephi, Cytoscape, and NetworkX to display nodes, edges, and their attributes in the map.High-risk nodes, risk transmission chains, and abnormal trading behaviors are highlighted through different colors, sizes, and shapes. Clustering algorithms are used to group risk nodes in the graph to form risk hotspots. High-risk areas are marked through heat maps, node size, and color depth. Front-end tools such as D3.js and Plotly are used to create interactive graph visualizations, allowing users to explore risk data in real time and view detailed information on each node and edge. Commodity trading risk reports are automatically generated based on commodity trading risk graphs and inference results.

[0035] Preferably, step S1 comprises the following steps: Step S11: Acquire commodity trading information data, wherein the commodity trading information data includes market data, logistics data, meteorological data and financial data; Step S12: preprocessing the commodity trading information data to generate standard commodity trading information data, wherein the data preprocessing includes data cleaning, data denoising, missing value filling and data standardization; Step S13: Perform commodity transaction external correlation analysis on the logistics data and meteorological data in the standard commodity transaction information data to generate commodity transaction external correlation data; Step S14: Based on the external correlation data of commodity transactions, the geographic-meteorological knowledge graph expansion technology is used to mark the special entity nodes of the logistics data and the meteorological data to obtain the special entity node data.

[0036] In the embodiment of the present invention, data including transaction price, transaction volume, information of both parties, market sentiment, etc. are obtained through channels such as real-time trading platforms and market analysis tools. Logistics data related to commodity transportation, storage, distribution, etc. are obtained, including transportation routes, transportation methods, cargo status, delivery time, etc., which can be obtained from logistics companies, transportation management systems, etc. Meteorological information affecting commodity transactions, including weather changes, natural disasters, climate conditions, etc., can usually be obtained through meteorological stations, satellite data, meteorological service platforms, etc. Financial status data of relevant companies or institutions, such as revenue, profit, cost, investment status, etc., are obtained. These data usually come from financial reports, financial databases, etc. Incomplete, repeated or abnormal data points are removed to ensure data quality. Signal processing methods (such as low-pass filtering, mean filtering, etc.) are used to clean up the noise in the data in order to extract more reliable information. Use appropriate filling methods (such as mean filling, interpolation, forward filling, etc.) to complete the missing data to avoid analysis bias due to missing values. Standardize heterogeneous data from different sources (such as normalization, Z-score standardization, etc.) to ensure that data can be compared and analyzed on a unified scale. Analyze the external correlation between logistics data, meteorological data and commodity transactions through statistical methods (such as Pearson correlation coefficient, covariance analysis, factor analysis, etc.), find external factors that affect commodity price fluctuations, generate data on the relationship between commodity transactions and external factors (such as weather, transportation efficiency, etc.), and analyze how these external factors affect the supply and demand relationship of commodities, price fluctuations, etc. Combine logistics data and meteorological data with geographic information systems (GIS) and meteorological knowledge graphs to identify important geographic entity nodes (such as ports, transportation routes, key meteorological factors, etc.). By marking these special entity nodes, node data with geographic and meteorological contexts is generated to provide more accurate basic data for subsequent analysis and decision-making.

[0037] Preferably, step S13 includes the following steps: Step S131: extracting key logistics features from the logistics data in the standard bulk commodity transaction information data, extracting the transportation distance, transportation method and delivery location of the bulk commodity transaction, and obtaining key logistics feature data; Step S132: extracting meteorological key features from the meteorological data in the standard commodity trading information data, extracting the temperature, precipitation and seasonal changes of commodity trading, and obtaining meteorological key feature data; Step S133: Calculate the Pearson correlation coefficient of the logistics key feature data and the meteorological key feature data, and construct an external correlation matrix based on the Pearson correlation coefficient to generate an external correlation matrix for bulk commodity transactions; Step S134: performing correlation strength analysis on the external correlation matrix of bulk commodity transactions to generate external correlation strength data; performing correlation feature extraction on the external correlation matrix of bulk commodity transactions based on the external correlation strength data to generate external correlation data of commodity transactions.

[0038] In an embodiment of the present invention, the geographic coordinates of the starting point and the destination point in the transportation process are extracted from the logistics data, and the distance between the two is calculated using a geographic information system (GIS). The Haversing formula or other geographic distance calculation formulas can be used to consider factors such as the curvature of the earth. The mode of transport field in the logistics data is analyzed to extract common modes of transport, such as land transport, sea transport, air transport, etc. The mode of transport field can be classified by natural language processing technology to ensure data consistency. Delivery location information, especially key delivery locations or ports, is extracted from the logistics data. These delivery location information are crucial for subsequent association analysis and can be standardized by place name recognition or direct use of address data. The extracted key feature data of logistics include transportation distance, mode of transport, and delivery location, providing a basis for subsequent analysis. Daily or monthly temperature data is extracted from meteorological data to calculate the average temperature, maximum temperature, minimum temperature, etc. of the region, with particular attention to seasonal trends related to commodity trading. Extract precipitation data to calculate the total precipitation in a certain area, or pay attention to the impact of precipitation changes on commodity transportation, especially interference with logistics. Analyze seasonal changes in meteorological data, such as periodic fluctuations in climate patterns. Seasonal meteorological characteristic data related to the season can be extracted through seasonal classification models (such as four seasons classification). By extracting meteorological data, the key meteorological characteristic data obtained include temperature, precipitation and seasonal changes, which provide support for subsequent correlation analysis. The Pearson correlation coefficient between the key characteristic data of logistics and the key characteristic data of meteorology is calculated to measure the strength of the linear relationship between them. The specific calculation method is as follows: ;in, and are the observed values ​​of logistics data and meteorological data, respectively. and is their mean, is the number of samples. This method calculates the linear relationship between two variables, ranging from -1 to 1, and the closer the value is to 1 or -1, the stronger the correlation is. Based on the Pearson correlation coefficient, the logistics key feature data and the meteorological key feature data are combined with the commodity trading data to construct an external correlation matrix. This matrix can show the correlation between different features and provide structured data for subsequent analysis. By analyzing the Pearson correlation coefficient in the external correlation matrix, the strength of the correlation between the features can be determined. For example, a threshold (such as 0.5) can be set to filter out features with strong correlation (positive or negative correlation), or the coefficients can be sorted to identify the most influential external factors. Through statistical analysis, a set of external correlation strength data is obtained to show the degree of correlation between the features, and these correlations can be visualized using charts (such as heat maps, correlation diagrams). Based on the external correlation strength data, the most relevant external features for commodity trading are extracted. These features can include transportation methods, climate conditions, etc., which directly or indirectly affect key factors such as commodity price fluctuations and supply chain efficiency.

[0039] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes: Step S21: Based on the special entity node data, time series modeling is performed on the standard bulk commodity trading information data to generate a dynamic time series graph of bulk commodity trading; node change rate is calculated on the dynamic time series graph of bulk commodity trading to obtain the node change rate of bulk commodity trading; Step S22: screening key nodes of the dynamic time series graph of commodity trading by the change rate of commodity trading nodes to obtain uncontrollable trading nodes; performing event perception on the dynamic time series graph of commodity trading according to the uncontrollable trading nodes to generate uncontrollable node events; Step S23: dynamically update the nodes and relationships of the dynamic time series graph of bulk commodity transactions according to the uncontrollable node events, and generate an event-driven bulk commodity transaction graph.

[0040] In an embodiment of the present invention, based on the special entity node data, the standard commodity trading information data is modeled in time sequence. The key to time series modeling is to model each event, node and its attributes in the commodity trading process as they change over time. Deep learning models such as dynamic graph technology, recursive neural network (RNN), long short-term memory (LSTM) and so on can be used to model trading behavior. Through each time node (for example, the date of the transaction, the time point of the event, etc.), a time series analysis of logistics, meteorology and other data is performed, and these features are associated with special entity nodes to form a time series data map. Combined with the market data of commodity transactions, a dynamic time series map model is constructed to represent the evolution of commodity transactions, where each node of the map represents a transaction entity or event, and the edge represents the relationship between entities. Calculate the rate of change of each node in the commodity trading map over time (i.e., the node change rate). The change rate can be calculated based on the time series change of node attributes. Common methods include: node change rate = (current node attribute value − node attribute value at the previous moment) ÷ node attribute value at the previous moment 100%, calculate the change rate of each node, analyze its change trend, and identify nodes with large changes for subsequent analysis. By calculating the change rate of the node, select nodes with more drastic dynamic changes as key nodes. The changes of these key nodes will significantly affect the process of commodity trading. For example, factors such as large changes in trading volume, logistics interruptions, and weather changes all affect key links in the trading process. Set a threshold for the change rate (for example, the change rate exceeds a certain value) to filter the nodes. According to business needs, you can select and filter out nodes with large changes as "uncontrollable trading nodes". Based on the results of node screening, identify uncontrollable trading nodes. These nodes are unpredictable or uncontrollable factors in transactions (such as natural disasters, political events, sudden market fluctuations, etc.), which will have a greater impact on commodity trading. For uncontrollable nodes, further analyze their impact on the commodity trading map and identify events related to these nodes (such as supply chain disruptions caused by climate change, changes in transportation routes, etc.). According to the changes of uncontrollable nodes, event-aware analysis is performed. This process involves monitoring the dynamic changes of uncontrollable nodes in the graph, combining external data sources (such as news, meteorological data, etc.) to identify and generate uncontrollable events in real time. Use event recognition technologies, such as text analysis, event detection algorithms, etc., to timely obtain and identify external events in commodity transactions that affect transactions (such as market emergencies or climate events). Update the attributes of these nodes based on the event information of uncontrollable nodes. For example, if weather changes cause a change in the mode of transportation of a logistics node, update the attributes of the node (such as transportation time, transportation volume, etc.). According to the changes of uncontrollable nodes, update the edges or relationships between related nodes. If a node of a transaction changes due to uncontrollable factors, the corresponding relationships (such as buyers and sellers, cargo flow, etc.) also need to be updated. Whenever an uncontrollable node event is triggered, the nodes and relationships in the graph are dynamically updated according to the event characteristics. This process can be achieved through automated scripts or rule-based triggers to ensure that the graph can reflect changes in the trading environment in real time. During the update process, the graph will present a more dynamic structure, reflecting the changes in trading relationships at different time points, thereby providing support for subsequent decision analysis.

[0041] Preferably, performing event perception on the dynamic time series graph of commodity trading based on the uncontrollable trading nodes includes: The external environment data is matched with the dynamic time series graph of commodity transactions according to the uncontrollable transaction nodes to obtain the external environment data; Trigger analysis of uncontrollable transaction nodes based on external environment data to obtain event trigger data; Perform dynamic event prediction on event triggering data to obtain dynamic event prediction data; Conduct market response analysis on dynamic event prediction data to obtain market response data; Based on the market response data, risk assessment and decision support are carried out on uncontrollable transaction nodes to obtain uncontrollable node events.

[0042] In an embodiment of the present invention, relevant external environmental data are obtained by matching with external environmental data sources based on uncontrollable transaction nodes. These environmental data include meteorological data, market data, regular changes, social events, etc. External data sources (such as meteorological stations, market monitoring systems, news media, etc.) are used to associate with nodes in the time series graph of bulk commodity transactions. External environmental data can be obtained in real time through interfaces or data capture tools, and data preprocessing is performed, including data cleaning, denoising, standardization, etc., to ensure the accuracy and consistency of the data. Geographic information system (GIS) or location recognition technology is used to match the geographical location information of bulk commodity transactions with external data such as meteorological data, logistics networks, and social events. The matched external environmental data is classified and sorted for further analysis and modeling. Association rule mining, pattern recognition and other methods are used to identify the trigger relationship between external environmental data and uncontrollable nodes. Machine learning models (such as decision trees, support vector machines, etc.) can be used to train trigger rules. For example, when the temperature changes sharply or the precipitation exceeds the threshold, the change of the logistics node is triggered, affecting the mode or time of transportation. The result of trigger analysis is event trigger data, which usually contains the following information: key factors in the external environment (such as weather, market changes, etc.), changes in uncontrollable trading nodes (such as logistics disruptions, changes in transportation routes, etc.). Use time series prediction models (such as LSTM neural networks, ARIMA models, etc.) to predict the time, intensity and impact of events. The prediction model should include responses to different external factors and predict future dynamic changes of uncontrollable trading nodes based on their historical behaviors. Market historical data can be used to analyze market reactions to similar events, and market response models can be established through regression analysis, market sentiment analysis, and other methods. By simulating market reactions in different scenarios (for example, logistics disruptions, price fluctuations, etc.), the market response to uncontrollable trading node events is evaluated. Risk analysis models (such as risk matrices, Monte Carlo simulations, etc.) are used in combination with market response data to quantify the risks of uncontrollable events. The evaluation involves multi-dimensional risk analysis, including financial risk, supply chain risk, market risk, etc. Based on the risk assessment results, decision support is provided to decision makers to help them make timely and reasonable adjustments. A risk decision report can be generated to provide suggestions such as adjusting trading strategies, optimizing logistics routes, and adjusting commodity pricing.

[0043] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes: Step S31: collecting event target interaction data on the event-driven commodity trading graph to obtain trading event interaction data; performing trading game analysis on the trading event interaction data to generate user trading game data; Step S32: constructing a transaction behavior profile for the transaction event interaction data according to the user transaction game data to generate an uncontrollable transaction behavior profile; using the uncontrollable transaction behavior profile to identify transaction behavior patterns of the event-driven commodity transaction graph to generate transaction behavior pattern data; Step S33: Generate an optimal trading decision path for the event-driven commodity trading graph through the trading behavior pattern data to obtain the optimal trading decision path; perform anomaly detection on the optimal trading decision path to generate optimal trading decision anomaly detection data; Step S34: Perform transaction risk transfer analysis based on the optimal transaction decision anomaly detection data to generate transaction risk transfer data; perform risk warning on the optimal transaction decision path based on the transaction risk transmission data to generate commodity transaction risk warning data.

[0044] In an embodiment of the present invention, by tracking and analyzing key events in the transaction process (such as price fluctuations, supply chain disruptions, natural disasters, etc.), the interaction data between the parties involved (such as suppliers, buyers, logistics parties, etc.) are collected. Data sources may include trading platforms, supply chain management systems, social media data, meteorological information, etc. Use data capture tools and API interfaces to connect with graph databases to obtain interaction data from different systems in an automated manner. Use game theory models (such as prisoner's dilemma, Nash equilibrium, etc.) to analyze how each participant makes decisions, collaborates or competes under event-driven conditions. According to the behavior and interaction patterns of different participants, evaluate the impact of the decisions of each party on market trends and transaction results. By clustering data such as transaction decisions, interaction methods, and transaction time, a behavioral profile of the participants is constructed. Use machine learning models (such as K-means clustering, decision trees, etc.) to classify transaction behaviors and form representative behavior patterns. Based on the uncontrollable transaction behavior portrait, transaction behavior pattern recognition is performed. This process helps understand the impact of uncontrollable factors on transaction behavior by identifying behavior patterns in different transaction scenarios. Use pattern recognition technology (such as neural networks, support vector machines, etc.) to extract regularities and anomalies in transaction behavior. Based on the identified patterns, analyze the impact of uncontrollable events (such as natural disasters, regular changes, etc.) on trading behavior. Generate the optimal trading decision path for event-driven commodity trading graphs through trading behavior pattern data. This process aims to generate the optimal trading path that adapts to market changes based on behavior patterns and external events. Use optimization algorithms (such as genetic algorithms, particle swarm optimization, etc.) to find the optimal trading decision path in the dynamically changing trading graph. The generated decision path includes the best time for trading execution, the best price range, the best logistics plan, etc. Perform anomaly detection on the optimal trading decision path. This process can help identify whether there are market fluctuations, external events or deviations in behavior patterns, which affect the effectiveness of the decision path. Use anomaly detection methods (such as Isolation Forest, density-based anomaly detection methods, etc.) to detect uncommon or unreasonable decisions in the trading path. Identify potential decision risks or problems in the execution process in a timely manner so as to adjust trading strategies. Perform trading risk transmission analysis based on the anomaly detection data of the optimal trading decision. This process analyzes how risks are transmitted in the trading chain and affect other trading nodes under abnormal circumstances. Use risk transfer models (such as Bayesian networks, system dynamics models, etc.) to simulate and quantify the risk transmission path. Analyze the key nodes and stages of risk transmission and identify potential risk accumulation points. Based on transaction risk transmission data, issue risk warnings for the optimal transaction decision path. Through the early warning system, timely issue transaction risk warnings to decision makers to remind them to take countermeasures. Real-time monitoring systems and rule engines can be used to combine historical data and real-time data to automatically trigger risk warnings.Early warning data includes the probability of occurrence of potential risks, the affected transaction nodes, and the risk mitigation measures that should be taken.

[0045] Preferably, using the uncontrollable transaction behavior profile to identify the transaction behavior pattern of the event-driven commodity transaction map includes: Extract commodity transaction price characteristics from the uncontrollable transaction behavior portrait to obtain commodity transaction price characteristic data; perform commodity transaction price fluctuation analysis on the commodity transaction price characteristic data through the relative strength index to generate commodity transaction price characteristic data; The commodity transaction price characteristic data is used to analyze the transaction volume of the uncontrollable transaction behavior portrait to generate the uncontrollable transaction volume; the commodity transaction price characteristic data and the uncontrollable transaction volume are used to identify the first transaction behavior pattern of the event-driven commodity transaction map to obtain the short-term manipulation behavior identification pattern; Confirm the price fundamentals of commodity transaction price feature data to obtain commodity price fundamentals data; perform deviation analysis on commodity transaction price feature data and commodity price fundamentals data to generate price fundamentals deviation data; use sentiment index to perform emotional evaluation on price fundamentals deviation data to generate commodity transaction emotional data; According to the commodity trading emotional data, the second trading behavior pattern is identified on the event-driven commodity trading map to obtain the emotional trading behavior identification pattern; the short-term manipulation behavior identification pattern and the emotional trading behavior identification pattern are integrated to obtain the trading behavior pattern data.

[0046] In the embodiment of the present invention, the price characteristics of commodity transactions are first extracted through the uncontrollable transaction behavior portrait. The purpose of this step is to extract key price variables from the transaction data to reflect the volatility, trend and market supply and demand relationship of commodity prices in the transaction process. Price characteristics include price range, historical price fluctuations, intraday fluctuations and other information. Statistical analysis and time series analysis methods (such as sliding window method and seasonal decomposition) are used to extract price fluctuation patterns. The relative strength index (RSI) is used to perform fluctuation analysis on commodity transaction price characteristic data. This step is used to determine whether the price is in an overbought or oversold state, thereby providing a basis for subsequent trading behavior prediction. The RSI indicator is used to quantify the intensity of price fluctuations and identify market reversal points. The potential impact of price fluctuations on trading decisions is analyzed, especially the extreme price fluctuations in the commodity market. Based on the commodity transaction price characteristic data, the uncontrollable transaction behavior portrait is further analyzed for trading volume. This step helps identify the relationship between changes in trading volume and price fluctuations, especially when prices fluctuate violently, trading volume will change abnormally, reflecting the activity of market participants. By calculating the correlation between trading volume and price fluctuations, the law of trading volume changes under specific price fluctuation scenarios is identified. Analyze market depth and liquidity to identify signs of short-term manipulation, such as abnormal volume increases and price increases. Based on commodity trading price feature data and uncontrollable trading volume, identify the first trading behavior pattern of event-driven commodity trading maps. This pattern focuses on identifying short-term manipulation, especially in the case of price fluctuations and abnormal trading volume. Use pattern recognition algorithms (such as decision trees, random forests, neural networks, etc.) to identify price manipulation and market-pulling behavior patterns. Determine the time and frequency of short-term manipulation and the trading nodes that are affected. Confirm the price fundamentals through the analysis of commodity trading price feature data. This step aims to confirm the impact of commodity fundamental information (such as supply chain conditions, production costs, market demand, etc.) on price fluctuations and help identify whether price fluctuations are driven by fundamental factors. Analyze the relationship between price and fundamental factors, especially factors such as changes in market supply and demand, producer costs, and seasonal demand. Use fundamental data to confirm the rationality or abnormality of price fluctuations. Evaluate whether prices deviate from the level of fundamental support by analyzing the deviation between price fundamental data and commodity trading price feature data. Calculate the deviation between prices and fundamental factors to identify when prices are overreacting, over-rising or falling. Price changes with large deviations indicate that market sentiment drivers are starting to take effect. Apply the sentiment index to perform an emotional assessment of price fundamental deviation data to generate commodity trading sentiment data. This assessment helps identify price fluctuations caused by market sentiment (such as panic, greed, etc.).Use sentiment analysis tools (such as social media sentiment analysis, news report sentiment analysis, etc.) to quantify market sentiment and correlate it with price fluctuations. Identify sentiment-driven price changes and provide data support for subsequent emotional trading behavior identification. Based on commodity trading sentiment data, conduct second trading behavior pattern identification based on event-driven commodity trading maps. This process mainly focuses on the identification of emotional trading behavior. By modeling the relationship between market sentiment and price fluctuations, identify trading patterns driven by emotions. For example, excessive buying behavior when market sentiment is too optimistic, or selling behavior when market sentiment is low. By integrating short-term manipulative behavior patterns and emotional trading behavior patterns, a comprehensive trading behavior pattern data is obtained. This model includes the interactive effects of manipulative behavior and emotional trading in the market, as well as their performance in different market scenarios. Integrate these two patterns to form a comprehensive trading behavior model to help analyze potential risks in the market and provide decision support for subsequent trading strategies.

[0047] Preferably, the transaction risk transfer analysis based on the optimal transaction decision anomaly detection data includes: Based on the abnormal detection data of the optimal trading decision, the trading risk transmission index is designed to obtain the key indicators of trading risk transmission, where the key indicators of trading risk transmission include risk exposure, transmission rate and risk concentration; the optimal trading decision abnormal detection data is matched with the indicator similarity using risk exposure, transmission rate and risk concentration to obtain the trading risk indicator matching data; Mapping the transaction risk indicator matching data to the event-driven commodity trading graph to construct the associated risk indicator path and generate transaction risk path data; marking the associated nodes of the transaction risk path data to obtain nodes with the same risk indicator; The propagation distance of nodes with the same risk index is calculated to obtain the node risk propagation distance data; the transaction risk transmission analysis is performed on the optimal transaction decision anomaly detection data through the node risk propagation distance data to generate transaction risk transmission data.

[0048] In the embodiment of the present invention, through 202410789791.4 For example, nodes with higher exposure are easily affected when prices fluctuate violently, and their trading decisions are greatly disturbed by external factors. Measures the speed at which risks spread from one node to other nodes. The transmission rate reflects the reaction speed of market participants and the efficiency of risk response measures. For example, nodes with faster risk transmission rates will accelerate price fluctuations in the market, or quickly expand to other related trading parties when risks occur. Measures the risk concentration of multiple nodes in the market. If multiple nodes have similar risk exposure or trading behaviors, they will share risks and form a risk concentration area. For example, multiple trading nodes of the same commodity are concentrated in a certain area or cycle and fluctuate, which is easy to lead to the phenomenon of risk concentration. The designed key indicators of transaction risk transmission include: risk exposure: the level of node exposure to potential risks. Transmission rate: the speed at which risks spread from one node to other nodes. Risk concentration: the concentration of risks at multiple nodes. The optimal transaction decision anomaly detection data is matched for indicator similarity using risk exposure, transmission rate and risk concentration. The purpose of this process is to find transaction decision nodes with similar risk characteristics and analyze their similarity relationships. Calculate the similarity between risk exposure, transmission rate and risk concentration through data analysis methods (such as Euclidean distance, cosine similarity or weighted matching measure). Similar nodes tend to behave in the same way during risk propagation. Identify trading nodes with high similarity in risk exposure, transmission rate and risk concentration according to the set threshold or standard. Map the matched trading risk indicator data to the event-driven commodity trading graph. In the graph, nodes represent trading behaviors or market participants, and edges represent trading relationships or risk transmission relationships between nodes. According to the trading risk indicator matching data, establish associated risk indicator paths, which reflect the potential paths for risk transmission from one node to other nodes. Identify potential risk transmission channels in the market by analyzing the associated paths in the graph. Mark the associated nodes of the trading risk path data and mark those nodes with the same risk indicators. These nodes are usually in the key position of risk transmission and are potential risk sources or transmission hubs. Mark the nodes with the same risk indicators to further analyze their impact on the entire trading system. The marked nodes are key nodes for risk transmission and will have a significant impact on trading decisions. Calculate the propagation distance between nodes with the same risk indicator. The propagation distance reflects the speed of risk transmission and the scope of influence between nodes. A shorter propagation distance means that the risk spreads quickly to other key nodes. The propagation distance between different nodes can be calculated through graph analysis algorithms (such as the shortest path algorithm and breadth-first search). Nodes with shorter propagation distances mean that they play a key role in the risk expansion process and can accelerate the spread of risks.The transaction risk transmission analysis is performed on the optimal transaction decision anomaly detection data based on the node risk propagation distance data. By analyzing the propagation path and propagation distance between nodes, the risk propagation mechanism in the market can be identified to help warn of potential market risks. Based on the propagation distance analysis results, the transaction risk transmission data is generated. This data includes potential risk propagation channels and key nodes in the market.

[0049] Preferably, step S4 comprises the following steps: Step S41: Integrate the commodity trading risk warning data into the event-driven commodity trading map to obtain the commodity trading risk map; set a query window for the commodity trading risk map to generate a commodity trading query window; Step S42: The commodity trading risk map is used to infer and visualize the commodity trading risk through the commodity trading query window, thereby generating a commodity trading risk report.

[0050] In an embodiment of the present invention, the commodity trading risk warning data (such as risk exposure, transmission rate, risk concentration, etc.) generated by the previous steps are integrated into an event-driven commodity trading graph. The nodes in the graph represent trading behaviors or market participants, and the edges represent the trading relationships and risk transmission paths between them. The warning data is integrated with the existing information in the graph (such as historical trading data, market behavior data, etc.) to ensure that each node and edge in the graph can reflect the current market risk situation. Through the graph enhancement method, a commodity trading risk graph containing trading risk warning information is created. The node status and edge weight of the graph are updated in combination with the risk data. The node status reflects the risk status of market participants in the current trading environment, and the edge weight reflects the risk transmission strength in the trading relationship. When the trading risk warning data changes, the relationship structure in the graph will also be adjusted accordingly, and the new risk information will affect the connection strength of the node and the speed of risk transmission in real time. In order to facilitate the analysis of the commodity trading risk graph, a query window is set, through which users can easily access and query the information in the graph. Users can select the query time range, trading risks of specific commodities, trading behaviors of specific market participants, etc. through the interface, and can customize queries through different parameters. Through the query window, users can infer the future trend and potential changes of trading risks based on the current state of the commodity trading risk map. Use graph reasoning algorithms (such as shortest path algorithm, Bayesian reasoning, etc.) to calculate the potential risk propagation path between nodes and infer trading risk events that occur in the current trading environment. The future risk exposure of different nodes in the market is obtained through reasoning, which helps identify market fluctuations, price changes and potential risk outbreak points. The results of trading risk reasoning are displayed through a graphical interface, allowing users to intuitively understand the current and future trading risk situation. After querying and visualizing the trading risk data, the system will automatically generate a commodity trading risk report, which describes in detail the potential risks in the market and their propagation paths, the risk impact of key trading behaviors, etc.

[0051] In this specification, a platform knowledge graph construction system based on commodity trading information is provided, which is used to execute the above-mentioned platform knowledge graph construction method based on commodity trading information. The platform knowledge graph construction system based on commodity trading information includes: The node setting module is used to obtain commodity trading information data; perform commodity trading external correlation analysis on the commodity trading information data to generate commodity trading external correlation data; perform special entity node marking on the commodity trading information data based on the commodity trading external correlation data to obtain special entity node data; The graph construction module is used to: perform time series modeling on commodity trading information data based on special entity node data to generate a dynamic time series graph of commodity trading; perform event perception on the dynamic time series graph of commodity trading to generate uncontrollable node events; dynamically update the graph nodes and relationships of the dynamic time series graph of commodity trading according to the uncontrollable node events to generate an event-driven commodity trading graph; The risk analysis module is used to collect event target interaction data for the event-driven commodity trading map to obtain transaction event interaction data; construct transaction behavior profiles through transaction event interaction data to generate uncontrollable transaction behavior profiles; use uncontrollable transaction behavior profiles to generate the optimal transaction decision path for the event-driven commodity trading map to obtain the optimal transaction decision path; perform transaction risk transmission analysis on the optimal transaction decision path to generate transaction risk transmission data; perform risk warning on the optimal transaction decision path based on the transaction risk transmission data to generate commodity trading risk warning data; The graph reasoning module is used to integrate commodity trading risk warning data into the event-driven commodity trading graph to obtain a commodity trading risk graph; perform commodity trading risk reasoning and visualization based on the commodity trading risk graph to generate a commodity trading risk report.

[0052] The beneficial effect of the present invention is that the node setting module provides detailed and accurate external correlation data for subsequent risk identification and graph construction by obtaining commodity trading information and conducting external correlation analysis of commodity transactions. This in-depth external correlation analysis can help identify external factors that affect transactions, such as market changes, logistics, weather, etc., thereby enhancing the accuracy of decision-making. The time series modeling function of the graph construction module can dynamically reflect the changes of various events in the commodity trading process and adjust the graph nodes and relationships in real time, which enables the system to perceive and update uncontrollable node events in real time during the transaction process, establish an event-driven graph model, and support dynamic decision-making. The risk analysis module can conduct in-depth analysis of the potential risks of transaction decisions by constructing transaction behavior portraits and analyzing the optimal transaction decision path, provide the most optimized transaction path selection, and conduct systematic early warning through transaction risk transmission analysis. The core of this module is to quantitatively analyze the transmission process of transaction risks, thereby providing early warning of emerging risks. The graph reasoning module integrates an event-driven commodity trading risk graph and provides a graph reasoning function, which supports the visualization of commodity trading risks through the reasoning process. This module not only provides deep data mining capabilities, but also can pass complex risk analysis processes to decision makers through visual reports, thereby improving the transparency and accuracy of decision-making. Therefore, the present invention solves the defects of the existing technology in commodity trading, which lacks dynamic updates, risk warnings, and accurate decision-making support, through dynamic time series modeling, event-driven analysis, and transaction behavior portrait construction, and improves the comprehensiveness and relevance of the knowledge graph.

[0053] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0054] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for constructing a platform knowledge graph based on commodity trading information, characterized in that: The following steps are involved: Step S1: Obtaining commodity transaction information data; performing commodity transaction external correlation analysis on the commodity transaction information data to generate commodity transaction external correlation data; Based on the external correlation data of commodity transactions, special entity node marking is performed on the commodity transaction information data to obtain special entity node data; Step S2: Based on the special entity node data, the commodity transaction information data is modeled in time series to generate a dynamic time series graph of commodity transactions; the dynamic time series graph of commodity transactions is event-aware to generate uncontrollable node events; Dynamically update the nodes and relationships of the dynamic time series graph of commodity transactions according to uncontrollable node events to generate an event-driven commodity transaction graph; Step S3: collecting event target interaction data on the event-driven commodity trading graph to obtain trading event interaction data; Build transaction behavior profiles through transaction event interaction data to generate uncontrollable transaction behavior profiles; Use the uncontrollable trading behavior profile to generate the optimal trading decision path for the event-driven commodity trading map to obtain the optimal trading decision path; Conduct transaction risk transfer analysis on the optimal transaction decision path and generate transaction risk transfer data; Provide risk warning for the optimal trading decision path based on transaction risk transmission data, and generate commodity trading risk warning data; Step S4: Integrate the commodity trading risk warning data into the event-driven commodity trading map to obtain the commodity trading risk map; Commodity trading risks are inferred and visualized based on the commodity trading risk map to generate a commodity trading risk report.

2. The method for constructing a platform knowledge graph based on commodity trading information according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: Acquire commodity trading information data, wherein the commodity trading information data includes market data, logistics data, meteorological data and financial data; Step S12: preprocessing the commodity trading information data to generate standard commodity trading information data, wherein the data preprocessing includes data cleaning, data denoising, missing value filling and data standardization; Step S13: Perform commodity transaction external correlation analysis on the logistics data and meteorological data in the standard commodity transaction information data to generate commodity transaction external correlation data; Step S14: Based on the external correlation data of commodity transactions, the geographic-meteorological knowledge graph expansion technology is used to mark the special entity nodes of the logistics data and the meteorological data to obtain the special entity node data.

3. The method for constructing a platform knowledge graph based on commodity trading information according to claim 2, characterized in that: Step S13 includes the following steps: Step S131: extracting key logistics features from the logistics data in the standard bulk commodity transaction information data, extracting the transportation distance, transportation method and delivery location of the bulk commodity transaction, and obtaining key logistics feature data; Step S132: extracting meteorological key features from the meteorological data in the standard commodity trading information data, extracting the temperature, precipitation and seasonal changes of commodity trading, and obtaining meteorological key feature data; Step S133: Calculate the Pearson correlation coefficient of the logistics key feature data and the meteorological key feature data, and construct an external correlation matrix based on the Pearson correlation coefficient to generate an external correlation matrix for bulk commodity transactions; Step S134: performing correlation strength analysis on the external correlation matrix of bulk commodity transactions to generate external correlation strength data; performing correlation feature extraction on the external correlation matrix of bulk commodity transactions based on the external correlation strength data to generate external correlation data of commodity transactions.

4. The method for constructing a platform knowledge graph based on commodity trading information according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: Based on the special entity node data, time series modeling is performed on the standard bulk commodity trading information data to generate a dynamic time series graph of bulk commodity trading; node change rate is calculated on the dynamic time series graph of bulk commodity trading to obtain the node change rate of bulk commodity trading; Step S22: screening key nodes of the dynamic time series graph of commodity trading by the change rate of commodity trading nodes to obtain uncontrollable trading nodes; performing event perception on the dynamic time series graph of commodity trading according to the uncontrollable trading nodes to generate uncontrollable node events; Step S23: dynamically update the nodes and relationships of the dynamic time series graph of bulk commodity transactions according to the uncontrollable node events, and generate an event-driven bulk commodity transaction graph.

5. The method for constructing a platform knowledge graph based on commodity trading information according to claim 4 is characterized in that: Event perception of the dynamic time series graph of commodity transactions based on uncontrollable transaction nodes includes: The external environment data is matched with the dynamic time series graph of commodity transactions according to the uncontrollable transaction nodes to obtain the external environment data; Trigger analysis of uncontrollable transaction nodes based on external environment data to obtain event trigger data; Perform dynamic event prediction on event triggering data to obtain dynamic event prediction data; Conduct market response analysis on dynamic event prediction data to obtain market response data; Based on the market response data, risk assessment and decision support are carried out on uncontrollable transaction nodes to obtain uncontrollable node events.

6. The method for constructing a platform knowledge graph based on commodity trading information according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: collecting event target interaction data on the event-driven commodity trading graph to obtain trading event interaction data; performing trading game analysis on the trading event interaction data to generate user trading game data; Step S32: constructing a transaction behavior profile for the transaction event interaction data according to the user transaction game data to generate an uncontrollable transaction behavior profile; using the uncontrollable transaction behavior profile to identify transaction behavior patterns of the event-driven commodity transaction graph to generate transaction behavior pattern data; Step S33: Generate an optimal trading decision path for the event-driven commodity trading graph through the trading behavior pattern data to obtain the optimal trading decision path; perform anomaly detection on the optimal trading decision path to generate optimal trading decision anomaly detection data; Step S34: Perform transaction risk transfer analysis based on the optimal transaction decision anomaly detection data to generate transaction risk transfer data; perform risk warning on the optimal transaction decision path based on the transaction risk transmission data to generate commodity transaction risk warning data.

7. The method for constructing a platform knowledge graph based on commodity trading information according to claim 1, characterized in that: Using uncontrollable trading behavior portraits to identify trading behavior patterns based on event-driven commodity trading graphs includes: Extract commodity transaction price characteristics from the uncontrollable transaction behavior portrait to obtain commodity transaction price characteristic data; perform commodity transaction price fluctuation analysis on the commodity transaction price characteristic data through the relative strength index to generate commodity transaction price characteristic data; The commodity transaction price characteristic data is used to analyze the transaction volume of the uncontrollable transaction behavior portrait to generate the uncontrollable transaction volume; the commodity transaction price characteristic data and the uncontrollable transaction volume are used to identify the first transaction behavior pattern of the event-driven commodity transaction map to obtain the short-term manipulation behavior identification pattern; Confirm the price fundamentals of commodity transaction price feature data to obtain commodity price fundamentals data; perform deviation analysis on commodity transaction price feature data and commodity price fundamentals data to generate price fundamentals deviation data; use sentiment index to perform emotional evaluation on price fundamentals deviation data to generate commodity transaction emotional data; According to the commodity trading emotional data, the second trading behavior pattern is identified on the event-driven commodity trading map to obtain the emotional trading behavior identification pattern; the short-term manipulation behavior identification pattern and the emotional trading behavior identification pattern are integrated to obtain the trading behavior pattern data.

8. The method for constructing a platform knowledge graph based on commodity trading information according to claim 6, characterized in that: Transaction risk transfer analysis based on optimal transaction decision anomaly detection data includes: Based on the abnormal detection data of the optimal trading decision, the trading risk transmission index is designed to obtain the key indicators of trading risk transmission, where the key indicators of trading risk transmission include risk exposure, transmission rate and risk concentration; the optimal trading decision abnormal detection data is matched with the indicator similarity using risk exposure, transmission rate and risk concentration to obtain the trading risk indicator matching data; Mapping the transaction risk indicator matching data to the event-driven commodity trading graph to construct the associated risk indicator path and generate transaction risk path data; marking the associated nodes of the transaction risk path data to obtain nodes with the same risk indicator; The propagation distance of nodes with the same risk index is calculated to obtain the node risk propagation distance data; the transaction risk transmission analysis is performed on the optimal transaction decision anomaly detection data through the node risk propagation distance data to generate transaction risk transmission data.

9. The method for constructing a platform knowledge graph based on commodity trading information according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: Integrate the commodity trading risk warning data into the event-driven commodity trading map to obtain the commodity trading risk map; set a query window for the commodity trading risk map to generate a commodity trading query window; Step S42: The commodity trading risk map is used to infer and visualize the commodity trading risk through the commodity trading query window, thereby generating a commodity trading risk report.

10. A platform knowledge graph construction system based on commodity trading information, characterized in that: Used to execute the platform knowledge graph construction method based on bulk commodity trading information as claimed in claim 1, the platform knowledge graph construction system based on bulk commodity trading information includes: The node setting module is used to obtain commodity trading information data; perform commodity trading external correlation analysis on the commodity trading information data to generate commodity trading external correlation data; perform special entity node marking on the commodity trading information data based on the commodity trading external correlation data to obtain special entity node data; The graph construction module is used to: perform time series modeling on commodity trading information data based on special entity node data to generate a dynamic time series graph of commodity trading; perform event perception on the dynamic time series graph of commodity trading to generate uncontrollable node events; dynamically update the graph nodes and relationships of the dynamic time series graph of commodity trading according to the uncontrollable node events to generate an event-driven commodity trading graph; The risk analysis module is used to collect event target interaction data for the event-driven commodity trading map to obtain transaction event interaction data; construct transaction behavior profiles through transaction event interaction data to generate uncontrollable transaction behavior profiles; use uncontrollable transaction behavior profiles to generate the optimal transaction decision path for the event-driven commodity trading map to obtain the optimal transaction decision path; perform transaction risk transmission analysis on the optimal transaction decision path to generate transaction risk transmission data; perform risk warning on the optimal transaction decision path based on the transaction risk transmission data to generate commodity trading risk warning data; The graph reasoning module is used to integrate commodity trading risk warning data into the event-driven commodity trading graph to obtain a commodity trading risk graph; perform commodity trading risk reasoning and visualization based on the commodity trading risk graph to generate a commodity trading risk report.

Images