Product security event evolution process prediction method and device, and electronic equipment

Abstract

The invention relates to a product security event evolution process prediction method and device, and electronic equipment. The method comprises the steps that a data set is built through multi-source heterogeneous data, and the multi-source heterogeneous data is data having an incidence relation with a product safety event; cleaning data and extracting event entities and attributes from the data set; constructing a affair graph by using a causal reasoning method of a time sequence association rule; taking entities in the affair atlas as nodes of a propagation network, and mapping causal relationships in the affair atlas as edges of the propagation network and weights on the edges; creating agent attributes and event state transition rules; and constructing a multi-agent model, and predicting the evolution process of the product security event by using the multi-agent model. According to the invention, the technical problem of lack of research on the evolution law of the product security event in the prior art is solved.

Description

Technical Field

[0001] This invention relates to the field of safety engineering technology, and in particular to a method, apparatus, and electronic device for predicting the evolution of product safety incidents. Background Technology

[0002] With the increasing variety of products and the globalization of supply chains, the scope and complexity of product safety incidents have significantly increased. Consumer feedback during actual product use often goes through multiple layers of processing. Although feedback is received immediately, the handling of related issues still involves a considerable delay. Furthermore, while consumer reports are widespread, different pieces of information often exist in isolation, lacking effective integration and analysis methods, making it difficult to conduct a comprehensive risk assessment of product safety issues.

[0003] Product safety issues not only directly threaten consumers' health and safety, but also quickly trigger widespread public attention and spread negative emotions. Current technology lacks research on the evolutionary patterns of product safety incidents. Summary of the Invention

[0004] To address the issue that product safety problems can quickly trigger widespread public attention and negative sentiment, and given the lack of existing technologies to study the evolution of product safety incidents, this invention provides a method, apparatus, and electronic device for predicting the evolution of product safety incidents. The technical solution is as follows:

[0005] On the one hand, a method for predicting the evolution of product safety incidents is provided, the method comprising: S1. Construct a dataset using multi-source heterogeneous data, wherein the multi-source heterogeneous data is data that is related to product security incidents; S2. Clean the data and extract event entities and attributes from the dataset; S3. Construct a causal reasoning graph using temporal association rules; S4. Use the entities in the event graph as nodes of the propagation network, and map the causal relationships in the event graph to the edges and weights of the propagation network. S5. Create agent attributes and event state transition rules; S6. Construct a multi-agent model and use the multi-agent model to predict the evolution of product safety events.

[0006] Optionally, S2 specifically includes: cleaning multi-source heterogeneous data, wherein the multi-source heterogeneous data includes: consumer complaint data and social media sentiment data; extracting event entities and attributes from consumer complaint data; performing sentiment analysis on social media sentiment data, and marking key nodes in the spread of product safety incidents.

[0007] Optionally, S3 specifically includes: importing the dataset into a graph database to construct a knowledge graph, wherein the entities in the knowledge graph include: event nodes, social media post nodes, and environmental variable nodes, and the relationships in the knowledge graph include: event triggering relationships and policy impact relationships; generating causal relationships between entities using a causal reasoning method based on temporal association rules; organizing the causal relationships into paths in a directed graph to generate causal chains; visualizing the knowledge graph and annotating the key paths in the causal chains to generate a reasoning graph.

[0008] Optionally, S4 specifically includes: using entities in the event graph as nodes of the propagation network, and mapping causal relationships in the event graph to edges and weights on the edges of the propagation network, wherein the weights are used to represent the influence strength between two nodes in the propagation network. S5 specifically includes: creating agent attributes, wherein the agent attributes include: the status, type, and scope of impact of product security events; and determining state transition rules; Specifically, S6 includes: constructing a multi-agent model based on the nodes, edges, and weights on the edges of the propagation network, agent attributes, and state transition rules.

[0009] Optionally, the state transition rules are determined, including: determining whether a product safety event changes from a latent state to an outbreak state based on the outbreak probability; and determining whether a product safety event changes from an outbreak state to a decay state based on the decay probability.

[0010] Optionally, the probability of outbreak and the probability of decline are determined in the following ways: in, and These are the probability of outbreak and the probability of decline, respectively. and These represent the base probability of a product safety incident and the base probability of its decline, respectively. The weights of each influencing factor, These are the normalized values ​​for each influencing factor. is the coupling coefficient.

[0011] Optionally, after using a multi-agent model to predict the evolution of product safety events, the method further includes: comparing the difference between the predicted values ​​of the multi-agent model and real historical data using mean squared error; calculating the structural similarity between the simulated propagation path of the product safety event and the historical propagation path; determining an intervention strategy based on the predicted values ​​of the multi-agent model, the intervention strategy being used to suppress the propagation of the product safety event; calculating the error gradient based on the predicted values ​​of the multi-agent model and real historical data, and using the error gradient to adjust the weights of the input variables of the multi-agent model in reverse.

[0012] On the other hand, a device for predicting the evolution of product safety incidents is provided, the device comprising: The first construction module is used to construct a dataset using multi-source heterogeneous data, wherein the multi-source heterogeneous data is data related to product safety events; clean the data and extract event entities and attributes from the dataset; The second construction module is used to construct a reasoning graph using a causal reasoning method based on temporal association rules; to use entities in the reasoning graph as nodes in a propagation network, and to map causal relationships in the reasoning graph to edges and weights on the edges of the propagation network; to create agent attributes and event state transition rules; and to construct a multi-agent model. The processing module is used to predict the evolution of product safety events using the multi-agent model.

[0013] On the other hand, an electronic device is provided, the device including: a processor; a memory storing computer-readable instructions, which, when executed by the processor, implement the above method.

[0014] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction is stored in the storage medium, and the at least one instruction is loaded and executed by a processor to implement the above method.

[0015] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: This invention achieves integrated analysis from causal mining to dynamic prediction by deeply integrating dynamic event graphs and multi-agent coupling models. Compared with traditional event evolution methods, it can perform more in-depth causal analysis, more accurate dynamic prediction, and more efficient decision support for product safety events, providing a complete intelligent solution for solving complex problems in the field of product safety. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of a method for predicting the evolution process of a product safety incident, provided in an embodiment of the present invention. Figure 2 This is a flowchart of a multi-source heterogeneous product security event data extraction method according to an embodiment of the present invention; Figure 3 This is a flowchart of a process for constructing a logic graph according to an embodiment of the present invention; Figure 4 This is a schematic diagram of a process for constructing a principle graph according to an embodiment of the present invention; Figure 5 This is a schematic diagram of a multi-agent model construction process according to an embodiment of the present invention; Figure 6 This is a block diagram illustrating a device for predicting the evolution of a product safety incident according to an exemplary embodiment; Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0018] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0019] In embodiments of the present invention, terms such as "exemplarily" and "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Rather, the use of the term "exemplary" is intended to present the concept in a specific manner. To make the technical problems, technical solutions, and advantages of the present invention clearer, a detailed description will be provided below in conjunction with the accompanying drawings and specific embodiments.

[0020] As mentioned above, product safety issues not only directly threaten consumers' health and safety, but also quickly trigger widespread public attention and the spread of negative emotions.

[0021] Existing technologies have attempted to utilize knowledge graphs to store product security incidents and their relationships, such as constructing static relationship networks between events, enterprises, and policies using graph databases like Neo4j. These technologies can visualize some causal relationships between events, but their causal weights are fixed and cannot be dynamically adjusted based on real-time data (such as public opinion trends). Furthermore, they only support rule-based explicit causal analysis and lack the ability to uncover potential transmission paths. These limitations are significant.

[0022] Some studies use complex networks (such as NetworkX) to simulate event propagation, treating events as nodes and relationships as edges, and predicting evolutionary trends using infectious disease models such as the SIR model (susceptible-infected-recovered). However, these studies have limitations: they only consider network topology, ignoring external driving factors such as environmental policies and corporate behavior; and parameters such as propagation probability require manual pre-setting and cannot be adaptively optimized.

[0023] For product safety incident data storage, traditional knowledge graph technology can only construct static causal relationship networks, with fixed causal weights that cannot be dynamically adjusted with data updates. For example, the impact of policy changes on an event needs to be manually set, leading to significant discrepancies between predicted and actual results. Therefore, this invention utilizes dynamic event graph technology to adjust the probability weights of causal relationships in real time.

[0024] For the dynamic evolution of events, existing complex network models only simulate event propagation based on topology, neglecting the interaction of multi-dimensional factors such as environmental policies, corporate responses, and public opinion diffusion. For example, they cannot quantify the synergistic effect of "policy tightening and public opinion fermentation" on the outbreak of an event. Therefore, this invention proposes a multi-agent coupling equation, incorporating environmental, behavioral, and informational factors into a unified computational framework.

[0025] For model validation and strategy optimization, existing systems rely on a single metric (such as prediction accuracy) to verify model effectiveness, and strategy optimization requires manual trial and error. For example, the evaluation cycle for the effectiveness of policy interventions can be as long as several weeks, making it impossible to respond quickly to dynamic changes. This invention constructs a multi-dimensional validation system that simultaneously evaluates mean squared error (MSE), network similarity (graph edit distance), and key node recall, enabling rapid response to dynamic changes.

[0026] Ultimately, this invention constructs a causal graph and uses causal reasoning algorithms to uncover event causes, triggering mechanisms, and chain reaction paths. It quantifies and analyzes the driving factors of event evolution from three dimensions: environmental variables, agent behavior, and information propagation. Utilizing the topological structure based on the causal graph, it identifies key nodes and positive / negative feedback mechanisms in the evolutionary process. By integrating the causal logic of the causal graph with complex network dynamics methods, a dynamic evolution model is designed to simulate the development trend of events under different intervention strategies. The effectiveness and predictive ability of the model are verified through historical case backtracking and multi-scenario simulations. This enables the study of the evolutionary patterns of product safety events. The technical solution proposed in this invention is described below with reference to specific embodiments.

[0027] This invention provides a method for predicting the evolution of product safety incidents. This method can be implemented by an electronic device, which can be a terminal or a server. Figure 1 The flowchart shown illustrates a method for predicting the evolution of product safety incidents. This method's processing flow may include the following steps:

[0028] Step S1: Construct a dataset using multi-source heterogeneous data, where the multi-source heterogeneous data is data that is related to product safety incidents.

[0029] Step S2: Clean the data and extract event entities and attributes from the dataset.

[0030] The above steps involve the acquisition, cleaning, and structured storage of multi-source heterogeneous data to provide high-quality data support for upper-level analysis. Multi-source heterogeneous data refers to data sets originating from different sources and with varying structures.

[0031] In an embodiment of the present invention, the Faker library of Python is used to generate a synthetic dataset containing consumer complaint text (CSV format), social media sentiment (JSON format), and government recall notices (PDF converted to text).

[0032] Step S3: Construct a causal reasoning graph using the temporal association rule method.

[0033] Event graphs are knowledge graphs used to represent and reason about causal and temporal relationships between events. They not only focus on static relationships between entities (such as "person A is person B's father" in a knowledge graph), but also emphasize dynamic logical relationships between events, such as "event A causes event B" or "event A occurs after event B".

[0034] Step S4: Use the entities in the event graph as nodes of the propagation network, and map the causal relationships in the event graph to the edges and weights of the propagation network. Step S5: Create agent attributes and event state transition rules; Step S6: Construct a multi-agent model and use the multi-agent model to predict the evolution of product safety events.

[0035] The above steps construct a multi-agent model. A multi-agent model (ABM) is a computational model that studies the behavior and evolution of complex systems by simulating the interactions between multiple autonomous "agents." Each agent has its own state, behavioral rules, and interaction methods with the environment or other agents.

[0036] The technical solution proposed in this invention achieves integrated analysis from causal mining to dynamic prediction through the deep fusion of dynamic event graph and multi-agent coupling model. Compared with traditional event evolution methods, this invention can perform more in-depth causal analysis, more accurate dynamic prediction and more efficient decision support for product safety events, providing a complete intelligent solution for solving complex problems in the field of product safety.

[0037] According to an optional embodiment of the present invention, step S2 includes the following steps: cleaning multi-source heterogeneous data, wherein the multi-source heterogeneous data includes: consumer complaint data and social media sentiment data; extracting event entities and attributes from consumer complaint data; performing sentiment analysis on social media sentiment data and marking key nodes in the spread of product safety incidents.

[0038] Figure 2 This is a flowchart illustrating a multi-source heterogeneous product security event data extraction method according to an embodiment of the present invention, such as... Figure 2 As shown, the multi-source heterogeneous data includes consumer complaint data and social media sentiment data. First, the dataset is cleaned to remove duplicates, missing values, outliers, and to standardize timestamps and geographic location information.

[0039] Removing duplicates refers to eliminating duplicate records to avoid statistical bias.

[0040] Missing value handling refers to deleting or filling missing fields to maintain data integrity.

[0041] Outlier handling refers to identifying and correcting unreasonable or erroneous data (such as future timestamps, illegal geographic coordinates, etc.).

[0042] Standardized timestamps and geographic locations refer to unified time formats and geographic codes (such as latitude and longitude) to facilitate cross-data source integration and spatiotemporal analysis.

[0043] The above technical solutions can improve data quality and ensure the accuracy and reliability of subsequent analysis results.

[0044] Then, using NLP tools (SnowNLP, Spacy) to extract event entities (such as "battery explosion" and "excessive food additives") and attributes (time, location, product model) from the complaint text, and to conduct sentiment analysis on social media public opinion, marking key nodes in the spread of the event (such as posts with more than 10,000 reposts).

[0045] The above technical solutions can help monitor brand reputation, identify potential crises, and provide support for public opinion early warning systems.

[0046] According to another optional embodiment of the present invention, step S3 includes the following steps: importing the dataset into a graph database to construct a knowledge graph, wherein the entities in the knowledge graph include: event nodes, social media post nodes, and environmental variable nodes, and the relationships in the knowledge graph include: event triggering relationships and policy impact relationships; generating causal relationships between entities using a causal reasoning method based on temporal association rules; organizing the causal relationships into paths in a directed graph to generate causal chains; visualizing the knowledge graph and labeling the key paths in the causal chains to generate a reasoning graph.

[0047] Figure 3 This is a flowchart of a process for constructing a logic graph, as provided in an embodiment of the present invention. Figure 3 As shown, py2neo is used to import preprocessed event data into Neo4j to build event nodes (events, products, locations, etc.), social media post nodes (products, complaint texts, etc.), environmental variable nodes (policies, policy implementation time, etc.) and relationships (event triggering: including probability weight, policy impact: including influence weight).

[0048] py2neo is a Python client library for interacting with the Neo4j graph database, and is well-suited for building, querying, and managing graph-based knowledge graphs such as reasoning graphs, semantic networks, social networks, etc.

[0049] Event nodes represent specific consumer complaints or product issues that have occurred, and their attributes include: time, location, event type, severity, etc.

[0050] Policy impact refers to the influence of a policy on a specific event or product, and the "influence weight" is used to measure the effectiveness of the policy's intervention on the event. For example, after the "Measures for the Administration of Defective Product Recalls" were issued, a certain brand accelerated the recall of problematic products.

[0051] Social media post nodes represent user posts, comments, etc., from social media platforms, and their attributes include: posting time, number of reposts, sentiment score, keywords, etc.

[0052] Environmental variable nodes represent external factors such as policies and regulations, economic indicators, and seasonal changes. Attributes include: policy name, implementation time, scope of application, and level of impact.

[0053] Event triggering relationship refers to the phenomenon where one event is caused by another event or external factor. The strength of the causal relationship between two events is represented by "probability weight". For example, a "battery explosion" event may be triggered by a "charger quality problem".

[0054] Figure 4 This is a schematic diagram of a process for constructing a reasoning graph according to an embodiment of the present invention, such as... Figure 4 As shown, based on the above nodes, a causal chain is generated using the causal reasoning method of temporal association rules. The event graph is visualized and key paths are marked to complete the construction of the event graph.

[0055] Temporal association rule mining refers to discovering frequently occurring event sequence patterns based on the time sequence of events. For example, it can discover the common pattern of "excessive food additives", "media exposure", and "regulatory intervention".

[0056] Causal reasoning methods utilize statistical correlation, Granger causality tests, Bayesian networks, and other techniques to infer potential causal relationships between events. They combine factors such as the temporal sequence of events, semantic similarity, and propagation paths to make a comprehensive judgment.

[0057] The inferred causal relationships are then organized into paths in a directed graph, forming "causal chains." The graph structure is displayed using the Neo4j browser or other graph visualization tools. Nodes are distinguished by different colors or shapes (e.g., events, products, policies). Weight values ​​(e.g., probability, influence) are displayed on edges to enhance readability. Causal paths with high weight, high frequency, or high influence are also marked. For example, the complete path of "a certain type of battery failure," "multiple fire accidents," and "national mandatory recall" can be marked.

[0058] The event graph construction scheme integrates event data, social media sentiment, and policy background information. Through graph database modeling and causal reasoning technology, it realizes the transformation from raw data to high-level knowledge.

[0059] According to some optional embodiments of the present invention, steps S4, S5, and S6 can be implemented by the following methods: using entities in the event graph as nodes of the propagation network, mapping causal relationships in the event graph to edges and weights on the edges of the propagation network, wherein the weights are used to represent the influence strength between two nodes in the propagation network; creating agent attributes, wherein the agent attributes include: the state, type, and scope of influence of the product safety event; determining state transition rules; and constructing a multi-agent model based on the nodes, edges, and weights on the edges of the propagation network, the agent attributes, and the state transition rules.

[0060] A multi-agent model (ABM) is a computational model that studies the behavior and evolution of complex systems by simulating the interactions between multiple autonomous "agents". Each agent has its own state, behavioral rules, and ways of interacting with the environment or other agents.

[0061] Figure 5 This is a schematic diagram of a multi-agent model construction process according to an embodiment of the present invention, such as... Figure 5 As shown, firstly, agent attributes are created, including event states (latency, outbreak, and decay), product type, and scope of influence. Then, state transition rules are defined: transition from latency to outbreak based on outbreak probability; transition from outbreak to decay based on decay probability. Finally, Python's NetworkX is used to map the causal relationships in the event graph to network edge weights, simulating the propagation process. This completes the construction of the multi-agent model.

[0062] Complex networks are irregular, nonlinear network structures with statistical characteristics, consisting of a large number of nodes and edges. They differ from regular networks (such as grids) or random graphs, and are closer to real-world network structures.

[0063] In embodiments of the present invention, a complex network structure is constructed using NetworkX, with entities (such as events, products, and locations) in the event graph as nodes; causal relationships and influence relationships in the graph as edges; and the weights on the edges represent the influence strength or propagation probability between two nodes.

[0064] The multi-agent model construction scheme proposed in this invention integrates event state modeling, network structure mapping, and propagation simulation mechanisms, possessing strong theoretical support and practical application potential. It can not only simulate the propagation path of events but also provide a powerful tool for risk assessment, crisis intervention, and policy effectiveness verification.

[0065] As some optional embodiments of the present invention, in the process of constructing the intelligent agent model, the state transition rules are determined, including: determining the transition of a product safety event from a latent state to an outbreak state based on the outbreak probability; and determining the transition of a product safety event from an outbreak state to a decay state based on the decay probability.

[0066] When certain external stimuli or activation of nearby nodes are met, product safety incidents enter an outbreak state with a certain probability. As time goes by or public attention diminishes, product safety incidents enter a decline state with a certain probability.

[0067] In product safety incident propagation models, propagation probability is used to quantify the likelihood that a product safety incident will transition from its current state (e.g., "latent period") to an active state (e.g., "outbreak period"). Within a unit time step, the probability that an event in a latent state (outbreak state) will transition to an outbreak state (decline state) due to the influence of various factors is called the outbreak probability (decline probability).

[0068] According to some optional embodiments of the present invention, the above-mentioned outbreak probability and decay probability are determined by the following methods: The technical solution proposed in this invention comprehensively considers factors such as neighbor status, event severity, social media popularity, and policy impact to calculate the propagation probability, wherein... and These are the probability of outbreak and the probability of decline, respectively. and These represent the base probability of a product safety incident and the base probability of its decline, respectively. The weights of each influencing factor, The normalized values ​​for each influencing factor (neighbor activity level, incident severity, social media buzz, policy impact), is the coupling coefficient.

[0069] Based on the above formula and product safety incident data, by controlling variables (such as adjusting policy intensity and social media popularity), we can verify which factors play a dominant role in the spread of the incident, and thus predict the spread of future incidents.

[0070] In some optional embodiments of the present invention, after predicting the evolution of a product safety event using a multi-agent model, the mean squared error is used to compare the difference between the predicted value of the multi-agent model and the actual historical data; and the structural similarity between the simulated propagation path of the product safety event and the historical propagation path is calculated; an intervention strategy is determined based on the predicted value of the multi-agent model, and the intervention strategy is used to suppress the propagation of the product safety event; the error gradient is calculated based on the predicted value of the multi-agent model and the actual historical data, and the weights of the input variables of the multi-agent model are adjusted in reverse using the error gradient.

[0071] In this embodiment, the mean squared error (MSE) is first used to compare simulated values ​​with historical data and calculate path similarity, quantifying the error between the predicted and actual values ​​of public opinion heat. Then, the intervention strategy is optimized based on the simulation results. Furthermore, the weights in the propagation probability are optimized using the backpropagation algorithm based on historical data, resulting in better model prediction performance.

[0072] After the model has a certain predictive ability, the next step is to use the simulation results to design and optimize intervention strategies in order to control the spread of events and enhance its application value in real-world scenarios.

[0073] The technical solution proposed in this invention adopts a four-layer progressive architecture, including a data layer, a knowledge layer, a model layer, and a verification layer. The data layer constructs a causal graph through the fusion of multi-source heterogeneous data. The knowledge layer combines causal reasoning algorithms to mine potential transmission paths between events. The model layer employs multi-agent modeling and complex network dynamics methods to quantify the coupling effects of environmental variables, agent behavior, and information propagation, dynamically simulating the event evolution process. The verification layer evaluates model accuracy through multi-dimensional indicators such as time-series prediction error, topological similarity, and key node identification accuracy.

[0074] The technical solution proposed in this invention has the following advantages compared with the prior art: First, by combining rule-based reasoning with algorithms such as Bayesian networks, it is possible not only to capture explicit causal relationships but also to discover potential, indirect transmission paths. This multi-layered causal analysis capability enables the system to more comprehensively understand the complex relationships between events, providing a more reliable knowledge base for subsequent prediction and intervention.

[0075] Secondly, regarding the accuracy of dynamic prediction, existing methods often employ static models or single-dimensional analysis, making it difficult to cope with the impact of sudden events and the coupling of multiple factors. This invention, by introducing multi-agent modeling and complex network dynamics, simulates the event evolution process as a dynamic system, fully considering the interaction of multiple factors such as environmental variables, agent behavior, and information propagation. This dynamic modeling method significantly improves the timeliness and accuracy of predictions, especially demonstrating stronger adaptability when dealing with sudden events.

[0076] Finally, regarding the efficiency of decision support, traditional systems rely on manual analysis and experience-based judgment, resulting in slow response times and difficulty in quantifying and evaluating the effectiveness of strategies. This invention, by constructing an automated strategy optimization framework combined with multi-objective optimization algorithms and virtual simulation capabilities, can rapidly generate and evaluate various intervention options. This not only significantly shortens the decision-making cycle but also ensures the scientific validity and operability of the strategies through quantitative indicators, achieving a shift from passive response to proactive regulation.

[0077] Compared with traditional event evolution methods, the technical solution proposed in this invention can perform more in-depth causal analysis, more accurate dynamic prediction, and more efficient decision support for product safety events, providing a complete intelligent solution for solving complex problems in the field of product safety.

[0078] Figure 6 This is a block diagram illustrating a device for predicting the evolution of a product safety incident according to an exemplary embodiment, with reference to... Figure 6 The device includes a first building module 610, a second building module 620, and a processing module 630, wherein... The first construction module 610 is used to construct a dataset using multi-source heterogeneous data, wherein the multi-source heterogeneous data is data that is related to product safety events; clean the data and extract event entities and attributes from the dataset; The second construction module 620 is used to construct a reasoning graph using a causal reasoning method based on temporal association rules; to use entities in the reasoning graph as nodes of a propagation network, and to map causal relationships in the reasoning graph as edges and weights on the edges of the propagation network; to create agent attributes and event state transition rules; and to construct a multi-agent model. Processing module 630 is used to predict the evolution of product safety events using a multi-agent model.

[0079] It should be noted that, Figure 6 Preferred embodiments of the illustrated examples can be found in the relevant descriptions of the method embodiments, and will not be repeated here.

[0080] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention, such as... Figure 7 As shown, the electronic device may include the above-mentioned Figure 6 The illustrated device predicts the evolution of a product safety incident. Optionally, the electronic device 710 may include a first processor 7001.

[0081] Optionally, the electronic device 710 may also include a memory 7002 and a transceiver 7003.

[0082] The first processor 7001, memory 7002, and transceiver 7003 can be connected via a communication bus.

[0083] The following is combined Figure 7 A detailed description of each component of the electronic device 710 is provided below: The first processor 7001 is the control center of the electronic device 710. It can be a single processor or a collective term for multiple processing elements. For example, the first processor 7001 can be one or more central processing units (CPUs), application-specific integrated circuits (ASICs), or one or more integrated circuits configured to implement embodiments of the present invention, such as one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs).

[0084] Optionally, the first processor 7001 can perform various functions of the electronic device 710 by running or executing software programs stored in the memory 7002 and calling data stored in the memory 7002.

[0085] In a specific implementation, as one example, the first processor 7001 may include one or more CPUs, for example... Figure 7 CPU0 and CPU1 are shown in the diagram.

[0086] In a specific implementation, as one example, the electronic device 710 may also include multiple processors, for example... Figure 7 The first processor 7001 and the second processor 7004 are shown in the diagram. Each of these processors can be a single-core processor or a multi-core processor. Here, "processor" can refer to one or more devices, circuits, and / or processing cores used to process data (e.g., computer program instructions).

[0087] The memory 7002 is used to store the software program that executes the present invention, and is controlled by the first processor 7001 to execute it. The specific implementation method can be referred to the above method embodiment, and will not be repeated here.

[0088] Optionally, the memory 7002 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 7002 may be integrated with the first processor 7001 or may exist independently and be accessed through the interface circuit of the electronic device 710. Figure 7 (Not shown in the image) is coupled to the first processor 7001, and this embodiment of the invention does not specifically limit this.

[0089] The transceiver 7003 is used to communicate with network devices or with terminal devices.

[0090] Alternatively, transceiver 7003 may include a receiver and a transmitter. Figure 7 (Not shown separately). The receiver is used to implement the receiving function, and the transmitter is used to implement the transmitting function.

[0091] Optionally, the transceiver 7003 can be integrated with the first processor 7001, or it can exist independently and be connected via the interface circuit of the electronic device 710. Figure 7 (Not shown in the image) is coupled to the first processor 7001, and this embodiment of the invention does not specifically limit this.

[0092] It should be noted that, Figure 7 The structure of the electronic device 710 shown does not constitute a limitation on the router. Actual knowledge structure identification devices may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0093] Furthermore, the technical effects of the electronic device 710 can be referenced from the technical effects of the prediction method for the evolution process of product safety events described in the above method embodiments, and will not be repeated here.

[0094] It should be understood that the first processor 7001 in this embodiment of the invention may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0095] It should also be understood that the memory 7002 in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0096] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0097] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0098] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0099] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0100] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0101] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0102] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0103] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0104] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for predicting the evolution of a product safety incident, characterized in that, include: S1. Construct a dataset using multi-source heterogeneous data, wherein the multi-source heterogeneous data is data that is related to product security incidents; S2. Clean the data and extract event entities and attributes from the dataset; S3. Construct a causal reasoning graph using temporal association rules; S4. Use the entities in the event graph as nodes of the propagation network, and map the causal relationships in the event graph to the edges and weights of the propagation network. S5. Create agent attributes and event state transition rules; S6. Construct a multi-agent model and use the multi-agent model to predict the evolution of product safety events.

2. The method according to claim 1, characterized in that, S2 specifically includes: Data cleaning is performed on the multi-source heterogeneous data, which includes consumer complaint data and social media sentiment data. Extract event entities and attributes from the consumer complaint data; Sentiment analysis was performed on the aforementioned social media sentiment data to identify key points in the spread of product safety incidents.

3. The method according to claim 1, characterized in that, S3 specifically includes: The dataset is imported into a graph database to construct a knowledge graph, wherein the entities in the knowledge graph include: event nodes, social media post nodes, and environmental variable nodes, and the relationships in the knowledge graph include: event triggering relationships and policy impact relationships. The causal relationships between the entities are generated using a causal reasoning method based on temporal association rules; The causal relationships are organized into paths in a directed graph to generate causal chains; The knowledge graph is visualized, and the key paths in the causal chain are labeled to generate the event graph.

4. The method according to claim 3, characterized in that, S4 specifically includes: The entities in the event graph are used as nodes of the propagation network, and the causal relationships in the event graph are mapped to the edges and weights on the edges of the propagation network, wherein the weights are used to represent the influence strength between two nodes in the propagation network. S5 specifically includes: Create agent attributes, wherein the agent attributes include: the state, type, and scope of impact of product security events; determine state transition rules; S6 specifically includes: The multi-agent model is constructed based on the nodes, edges, and weights on the edges of the propagation network, the agent attributes, and the state transition rules.

5. The method according to claim 4, characterized in that, The determination of state transition rules includes: Based on the outbreak probability, it is determined that the product safety event has changed from a latent state to an outbreak state; Based on the decay probability, it is determined that the product safety event has changed from an outbreak state to a decay state.

6. The method according to claim 5, characterized in that, The outbreak probability and the decay probability are determined respectively by the following methods: in, and These are the outbreak probability and the decay probability, respectively. and These are the base probability of an outbreak and the base probability of a decay of the product safety incident, respectively. The weights of each influencing factor, These are the normalized values ​​for each influencing factor. is the coupling coefficient.

7. The method according to claim 1, characterized in that, After using the multi-agent model to predict the evolution of product safety events, the method further includes: The mean squared error is used to compare the difference between the predicted values ​​of the multi-agent model and the actual historical data; and the structural similarity between the simulated propagation path of the product safety event and the historical propagation path is calculated. An intervention strategy is determined based on the predicted values ​​from the multi-agent model, and the intervention strategy is used to suppress the spread of the product safety incident. The error gradient is calculated based on the predicted values ​​of the multi-agent model and the real historical data, and the weights of the input variables of the multi-agent model are adjusted in reverse using the error gradient.

8. A device for predicting the evolution of a product safety incident, the device being used to implement the method according to any one of claims 1-7, characterized in that, The device includes: The first construction module is used to construct a dataset using multi-source heterogeneous data, wherein the multi-source heterogeneous data is data related to product safety events; clean the data and extract event entities and attributes from the dataset; The second construction module is used to construct a reasoning graph using a causal reasoning method based on temporal association rules; to use entities in the reasoning graph as nodes in a propagation network, and to map causal relationships in the reasoning graph to edges and weights on the edges of the propagation network; to create agent attributes and event state transition rules; and to construct a multi-agent model. The processing module is used to predict the evolution of product safety events using the multi-agent model.

9. An electronic device, characterized in that, include: processor; A memory storing computer-readable instructions that, when executed by the processor, implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium contains program code that can be invoked by a processor to execute the method as described in any one of claims 1 to 7.

Images