Management decision-making method and system based on knowledge base construction technology

Abstract

The invention discloses a management decision-making method and system based on a knowledge base construction technology. The method comprises the following steps: performing sequential relationship extraction on multi-source financial data to obtain a sequential relationship set related to query content; constructing an event-entity incidence matrix corresponding to the time sequence relation set; according to the time sequence relation set and the event-entity incidence matrix, constructing a dynamic knowledge graph; determining causal effect parameters in the causal graph structure by adopting a dual machine learning model; constructing a structural causal model according to the causal graph structure and the causal effect parameters; and generating an anti-fact prediction result by using the structural causal model, and generating a decision scheme corresponding to the query content based on the anti-fact prediction result. The technical problem that decision information including accurate causal basis and prospective simulation information cannot be generated due to the fact that the causal relationship between financial data is difficult to determine and the intervention effect cannot be dynamically deduced in a related management decision method is solved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and more specifically, to a management decision-making method and system based on knowledge base construction technology. Background Technology

[0002] Current financial decision support methods primarily rely on statistical analysis and correlation mining of historical data, making it difficult to effectively distinguish between causal and correlational relationships between variables. Due to a lack of in-depth modeling of causal mechanisms within dynamic financial systems, these methods are unable to perform counterfactual analysis and effect prediction for various intervention policies, resulting in decision-making schemes lacking accurate causal basis and forward-looking simulation support.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This application provides a management decision-making method and system based on knowledge base construction technology, which at least solves the technical problem in relevant management decision-making methods that it is difficult to determine the causal relationship between financial data and cannot dynamically extrapolate the intervention effect, resulting in the inability to generate decision information including accurate causal evidence and forward-looking simulation information.

[0005] According to one aspect of this application, a management decision-making method based on knowledge base construction technology is provided, comprising: receiving query content; extracting time-series relationships from multi-source financial data based on the query content to obtain a set of time-series relationships related to the query content, wherein the set of time-series relationships is a structured data set used to dynamically represent the relationships between financial entities; constructing an event-entity association matrix corresponding to the set of time-series relationships, wherein the event-entity association matrix is ​​used to quantify the association strength between financial events and financial entities; constructing a dynamic knowledge graph based on the set of time-series relationships and the event-entity association matrix, and using a hybrid causal discovery method to identify the causes corresponding to financial variables in the query content in the dynamic knowledge graph. The causal graph structure, comprising a hybrid causal discovery method including constraint-based and score-based causal discovery algorithms, provides a graphical representation of the causal relationships and transmission paths between financial variables. A dual machine learning model is employed to determine causal effect parameters within the causal graph structure, which quantify the causal impact of interventions targeting the query content on the target variable. Based on the causal graph structure and causal effect parameters, a structural causal model is constructed, including structural equations defining the causal mechanisms between financial variables. The structural causal model is then used to generate counterfactual prediction results, and based on these results, a decision-making scheme corresponding to the query content is generated.

[0006] Optionally, a dynamic knowledge graph is constructed based on the temporal relation set and the event-entity association matrix, including: constructing an initial graph structure with time stamps based on the subject, object, relation, and timestamp information in the temporal relation set, wherein the timestamp information is used to identify the start time or effective time period of the relation; aligning the nodes in the initial graph structure based on the similarity of entity attributes to obtain a baseline graph; adding a time interval attribute to each relation edge in the baseline graph based on the timestamp information in the temporal relation set, wherein the time interval attribute is jointly defined by the start and end timestamps of the relation; and based on the predefined confidence level in the temporal relation set or based on the common... The statistical characteristics of current frequency are used to calculate the relationship strength weight for each relation edge. The baseline graph carrying the relationship strength weight and time interval attribute is determined as the main graph structure. Event feature vectors are generated from the entity nodes in the main graph structure using the event-entity association matrix. The event feature vectors are used as the initial features of the nodes using a time-series graph embedding algorithm. At the same time, the topological connections, relationship strength, and time interval in the main graph structure are learned to obtain the graph embedding representation. The baseline graph and the graph embedding representation obtained by the time-series graph embedding algorithm are stored together in the graph database to obtain a dynamic knowledge graph. The graph embedding representation is used as a vectorized attribute of the graph node and is stored in association with the topological structure of the baseline graph.

[0007] Optionally, a hybrid causal discovery method is employed to identify the causal graph structure corresponding to financial variables in the query content within a dynamic knowledge graph. This includes: extracting a set of core financial variables related to the query intent of the query content from the dynamic knowledge graph and obtaining the graph embedding representation corresponding to the core financial variables; constructing an initial causal skeleton graph using a constraint-based causal discovery algorithm, with the set of core financial variables as nodes and the time priority constraints included in the time sequence relation set in the dynamic knowledge graph as necessary conditions for determining the causal direction, wherein the initial causal skeleton graph is used to characterize whether there is a causal relationship between variables; and, based on the initial causal skeleton graph, employing a constraint-based causal discovery algorithm... The causal discovery algorithm for scores incorporates the similarity between graph embedding representations into the graph structure scoring function and uses key event association edges in the event-entity association matrix that exceed a preset threshold as directional constraints to strengthen the search, thereby optimizing the candidate causal graph structure. Based on the timestamp information in the temporal relationship set, causal edges in the candidate causal graph structure that do not conform to the preset temporal logic are removed, and the confidence score of the retained causal edges is calculated using historical data in the dynamic knowledge graph. Based on the temporally verified causal edges and their confidence scores, a causal graph structure is generated, which includes temporal lag attributes and confidence weights.

[0008] Optionally, a dual machine learning model is employed to determine the causal effect parameters in the causal graph structure, including: determining the input feature set for causal inference based on the causal path in the causal graph structure, wherein the input feature set includes at least: a treatment variable, an outcome variable, and a set of confounding variables, where the treatment variable represents the intervention to be implemented, the outcome variable represents the target indicator to be evaluated, and the set of confounding variables controls for background factors that jointly affect the treatment variable and the outcome variable; using a first machine learning model, with the set of confounding variables as input, predicting the treatment variable to obtain a first predicted value of the treatment variable and calculating a first residual between the actual value of the treatment variable and the first predicted value; using a second machine learning model, with the set of confounding variables as input, predicting the outcome variable to obtain a first residual between the actual value of the treatment variable and the first predicted value. The second predicted value is used to calculate the second residual between the actual value of the outcome variable and the second predicted value. The second machine learning model and the first machine learning model are independent of each other in terms of model structure or model parameters. The correlation between the first residual and the second residual is calculated using orthogonalization techniques to obtain the average treatment effect parameter. The average treatment effect parameter is used to characterize the global average causal effect of the treatment variable on the outcome variable. Based on the heterogeneity analysis model, the conditional average treatment effect parameter is calculated for different subgroups using the confounding variable set as the conditional variable. The heterogeneity analysis model is a model based on decision trees or meta-learners, and the different subgroups are sample sets with similar characteristics divided according to the feature space of the confounding variables. The average treatment effect parameter and the conditional average treatment effect parameter are jointly determined as the causal effect parameter.

[0009] Optionally, a structural causal model is constructed based on the causal graph structure and causal effect parameters, including: constructing a structural equation framework based on the nodes and edges in the causal graph structure, wherein each structural equation in the structural equation framework adopts a linear function form, representing the current variable as a linear combination of all parent variables of the current variable in the causal graph structure; based on the causal effect parameters, the initial estimated values ​​of each path coefficient in the structural equation framework are inversely estimated by calculating the path coefficients based on the structural equation model, thereby obtaining the initial structural equation; based on historical data in the dynamic knowledge graph, the maximum likelihood estimation method is used to perform joint parameter estimation on the initial structural equation, thereby obtaining an optimized structural equation parameter set, wherein the optimized structural equation parameter set includes: point estimates of path coefficients and confidence intervals of path coefficients; and configuring the optimized structural equation parameter set onto the corresponding edges of the causal graph structure to obtain the structural causal model.

[0010] Optionally, generating counterfactual prediction results using a structural causal model includes: determining a set of intervention operations based on query content, wherein each intervention operation in the set is represented by a preset operator, which is used to simulate intervention on preset processing variables in the causal graph structure; for each intervention operation in the set, calculating the potential results of the intervention operation on the outcome variable based on the optimized structural equation in the structural causal model to obtain a basic counterfactual prediction; quantifying the uncertainty of the basic counterfactual prediction based on the confidence intervals of the path coefficients in the optimized structural equation parameter set to obtain a counterfactual prediction distribution including the confidence intervals; and integrating the counterfactual prediction distributions corresponding to all intervention operations in the set to obtain the counterfactual prediction result.

[0011] Optionally, based on the counterfactual prediction results, a decision scheme corresponding to the query content is generated, including: determining the optimization objective and constraint set based on the business objectives of the query content, wherein the optimization objective includes at least: maximizing revenue and minimizing risk, and the constraint set comes from business rules and risk limits in the dynamic knowledge graph; establishing a multi-objective optimization function based on the expected effect value in the counterfactual prediction results as the basis for revenue estimation and the confidence interval in the counterfactual prediction results as the basis for risk assessment; constructing risk constraints based on the risk transmission paths identified in the causal graph structure and adding the risk constraints to the constraint set; solving the multi-objective optimization function using a multi-objective optimization algorithm to obtain the optimal solution set, wherein each solution in the optimal solution set corresponds to a candidate decision strategy and the expected performance of the candidate decision strategy on each optimization objective; calculating the robustness score of each strategy based on the causal path length and node influence range of the candidate decision strategy in the causal graph structure; selecting the optimal decision strategy from the optimal solution set according to the robustness score, and transforming the optimal decision strategy into an executable decision scheme.

[0012] According to another aspect of this application, a management decision-making system based on knowledge base construction technology is also provided, comprising: a receiving module, configured to receive query content, and extract time-series relationships from multi-source financial data based on the query content to obtain a set of time-series relationships related to the query content, wherein the set of time-series relationships is a structured data set used to dynamically represent the relationships between financial entities; a first construction module, configured to construct an event-entity association matrix corresponding to the set of time-series relationships, wherein the event-entity association matrix is ​​used to quantify the association strength between financial events and financial entities; and a second construction module, configured to construct a dynamic knowledge graph based on the set of time-series relationships and the event-entity association matrix, and use a hybrid causal discovery method to identify the financial variables corresponding to the query content in the dynamic knowledge graph. The system comprises a causal graph structure, wherein the hybrid causal discovery method includes a constraint-based causal discovery algorithm and a score-based causal discovery algorithm. The causal graph structure is a graphical representation of the causal relationships and transmission paths between financial variables. A determination module is used to determine the causal effect parameters in the causal graph structure using a dual machine learning model. These causal effect parameters are used to quantify the causal impact of interventions targeting the query content on the target variable. A third construction module is used to construct a structural causal model based on the causal graph structure and the causal effect parameters. This structural causal model includes structural equations defining the causal mechanisms between financial variables. A generation module is used to generate counterfactual prediction results using the structural causal model and, based on these counterfactual prediction results, generate decision schemes corresponding to the query content.

[0013] According to another aspect of this application, a non-volatile storage medium is also provided, the storage medium including a stored program, wherein the program, when running, controls the device where the storage medium is located to execute the above-mentioned management decision-making method based on knowledge base construction technology.

[0014] According to another aspect of this application, an electronic device is also provided, comprising: a memory and a processor, the processor being configured to run a program stored in the memory, wherein the program, when running, executes the above-described management decision-making method based on knowledge base construction technology.

[0015] According to another aspect of this application, a computer program is also provided, wherein when the computer program is executed by a processor, it implements the above-described management decision-making method based on knowledge base construction technology.

[0016] According to another aspect of this application, a computer program product is also provided, the computer program product including a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and the computer program, when executed by a processor, implements the above-mentioned management decision-making method based on knowledge base construction technology.

[0017] In this application, a query is received, and based on the query, temporal relationships are extracted from multi-source financial data to obtain a set of temporal relationships related to the query. This set of temporal relationships is a structured data set used to dynamically represent the relationships between financial entities. An event-entity association matrix corresponding to the temporal relationship set is constructed, whereby the event-entity association matrix is ​​used to quantify the association strength between financial events and financial entities. Based on the temporal relationship set and the event-entity association matrix, a dynamic knowledge graph is constructed, and a hybrid causal discovery method is used to identify the causal graph structure corresponding to the financial variables in the query within the dynamic knowledge graph. This hybrid causal discovery method includes a constraint-based causal discovery algorithm and a score-based causal discovery algorithm. The causal graph structure is a graphical representation of the causal relationships and transmission paths between financial variables. A dual machine learning model is employed to confirm... The causal effect parameters in the causal graph structure are used to quantify the causal impact of intervention measures targeting the query content on the target variable. Based on the causal graph structure and causal effect parameters, a structural causal model is constructed, which includes structural equations to define the causal mechanism between financial variables. The structural causal model is used to generate counterfactual prediction results, and based on these results, decision-making schemes corresponding to the query content are generated. This achieves the goal of determining the causal relationship between financial data and dynamically extrapolating the intervention effect, thus realizing the technical effect of generating decision information including accurate causal evidence and forward-looking simulation information. This solves the technical problem in relevant management decision-making methods where it is difficult to determine the causal relationship between financial data and to dynamically extrapolate the intervention effect, resulting in the inability to generate decision information including accurate causal evidence and forward-looking simulation information. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0019] Figure 1 This is a flowchart of a management decision-making method based on knowledge base construction technology according to an embodiment of this application;

[0020] Figure 2 This is a structural diagram of a management decision-making system based on knowledge base construction technology according to an embodiment of this application;

[0021] Figure 3 This is a hardware structure block diagram of a computer terminal for a management decision-making method based on knowledge base construction technology according to an embodiment of this application. Detailed Implementation

[0022] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0023] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0024] According to an embodiment of this application, a method embodiment of a management decision-making method based on knowledge base construction technology is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0025] Figure 1 This is a flowchart of a management decision-making method based on knowledge base construction technology according to an embodiment of this application, such as... Figure 1 As shown, the method includes the following steps:

[0026] Step S101: Receive query content, extract time-series relationships from multi-source financial data based on the query content, and obtain a time-series relationship set related to the query content. The time-series relationship set is a structured data set used to dynamically represent the relationship between financial entities.

[0027] In step S101, a query request input by the user in natural language is received. The query intent is analyzed using semantic understanding technology, and then temporal relationships related to the query content are extracted from multi-source financial data. Specifically, a relation extraction method based on a pre-trained language model can be used to identify financial entities and their relationships from unstructured texts such as financial news, company announcements, and research reports. Each relationship is labeled with a timestamp and confidence level, forming a set of temporal relationships composed of elements such as subject, relationship, object, and timestamp, thereby dynamically recording the evolution of the relationships between financial entities.

[0028] Step S102: Construct the event-entity association matrix corresponding to the time-series relationship set, wherein the event-entity association matrix is ​​used to quantify the association strength between financial events and financial entities.

[0029] In step S102, a standardized financial event classification system is first established, including types such as monetary policy adjustments, financial report releases, and mergers and acquisitions. Then, by calculating indicators such as the co-occurrence frequency, semantic similarity, and degree of impact between events and entities, the strength of their association is comprehensively evaluated, and finally, an association matrix is ​​formed with events as rows and entities as columns, where each matrix element represents the quantitative degree of association between a specific event and an entity.

[0030] Step S103: Based on the time sequence relationship set and the event-entity association matrix, a dynamic knowledge graph is constructed, and a hybrid causal discovery method is used to identify the causal graph structure corresponding to the financial variables in the query content in the dynamic knowledge graph. The hybrid causal discovery method includes a constraint-based causal discovery algorithm and a score-based causal discovery algorithm. The causal graph structure is a graphical representation of the causal relationship and transmission path between financial variables.

[0031] In step S103, the temporal relationship set is fused with the event-entity association matrix to construct a dynamic knowledge graph that supports temporal queries. This includes steps such as entity alignment, relationship disambiguation, and graph embedding learning, forming a comprehensive knowledge representation that incorporates topological structure, temporal dimension, and semantic information. Based on this dynamic knowledge graph, a hybrid causal discovery method is used to identify the causal graph structure between financial variables: first, a constraint-based causal discovery algorithm is used to construct an initial causal skeleton to determine the basic relationships between variables; then, a score-based causal discovery algorithm is combined for structural optimization. By optimizing the graph structure scoring function and introducing time priority constraints, a causal graph with a clear causal direction and transmission path is generated.

[0032] Step S104: Using a dual machine learning model, determine the causal effect parameters in the causal graph structure. The causal effect parameters are used to quantitatively evaluate the degree of causal influence of intervention measures targeting the query content on the target variable.

[0033] In step S104, a dual machine learning model is used to estimate the causal effect parameters. The dual machine learning model ensures the accuracy of the estimation through two stages: the first stage uses the machine learning model to predict the treatment variable and the outcome variable respectively, and the second stage calculates the causal effect based on the residuals. This method can effectively control the influence of confounding variables and accurately estimate the average treatment effect and the conditional average treatment effect.

[0034] Step S105: Based on the causal graph structure and causal effect parameters, construct a structural causal model, wherein the structural causal model includes structural equations for defining the causal mechanism between financial variables.

[0035] In step S105, based on the obtained causal graph structure and causal effect parameters, a structural causal model is constructed to formally represent the causal mechanism between financial variables. The structural causal model consists of a set of structural equations, each describing the functional relationship between a variable and its direct cause variable. The equation parameters are initialized by the causal effect parameters and optimized using the maximum likelihood estimation method to form a complete mathematical representation of the causal system.

[0036] Step S106: Use the structural causal model to generate counterfactual prediction results, and based on the counterfactual prediction results, generate decision schemes corresponding to the query content.

[0037] In step S106, counterfactual reasoning is performed using a structural causal model to simulate potential outcomes under different decision-making scenarios through intervention operations. Specifically, do-calculus and Monte Carlo simulation methods can be used to generate a counterfactual prediction distribution including confidence intervals. Based on the counterfactual prediction distribution, a multi-objective optimization model is established, comprehensively considering multiple factors such as benefits, risks, and constraints. The Pareto optimal solution set is solved through optimization algorithms, and the robustness of each scheme is evaluated. Finally, an optimal decision scheme with causal basis and forward-looking simulation support is generated, completing the intelligent transformation from data to decision.

[0038] Optionally, the decision-making scheme can involve controlling different agents to perform different operations. For example, when generating the decision-making scheme, different intervention operations are mapped to corresponding business agent execution units, such as allocating interest rate adjustment instructions to the business agent of a financial institution, issuing credit policies to the agent of a financial institution, and allocating investment strategies to the agent of an asset management institution. The instructions received by each agent are personalized based on counterfactual prediction results. By analyzing the transmission path and scope of influence of different intervention measures in the causal network, the optimal operating parameters and execution timing of each agent are determined.

[0039] Specifically, in the decision-making process, a multi-agent collaborative optimization model is first established based on counterfactual prediction results. This model comprehensively considers the interactions between different agent operations, analyzing the synergistic effects and conflict risks among operations using a causal graph structure to ensure that the operations of multiple agents can form a combined force rather than cancel each other out. For example, when a conflict is predicted between the interest rate hike by the monetary policy agent and the tax reduction measures by the fiscal policy agent, the operational parameters and execution order are automatically adjusted to optimize the overall effect.

[0040] Furthermore, the specific operational plan generated for each agent includes complete execution elements: clear operational instructions, expected performance indicators, execution time windows, necessary risk control measures, and coordination requirements with other agents. These plans are transmitted to the corresponding business systems through standardized interfaces. The actual performance data generated by the agents during execution is then fed back to the business systems to verify the accuracy of the causal model and optimize subsequent decisions. This multi-agent decision-making mechanism based on causal reasoning enables various components of the financial system to operate collaboratively based on a unified causal understanding, significantly improving decision-making efficiency and execution effectiveness in complex financial environments.

[0041] Through the above steps S101 to S106, the purpose of determining the causal relationship between financial data and dynamically inferring the intervention effect is achieved, thereby realizing the technical effect of generating decision information including accurate causal basis and forward-looking simulation information.

[0042] The following are Figure 1 The steps shown are illustrated and explained by way of example.

[0043] According to some optional embodiments of this application, a dynamic knowledge graph can be constructed based on a temporal relation set and an event-entity association matrix, which can be achieved through the following methods: Based on the subject, object, relation, and timestamp information in the temporal relation set, an initial graph structure carrying time stamps is constructed, wherein the timestamp information is used to identify the start time or effective time period of the relation; based on the similarity of entity attributes, the nodes in the initial graph structure are aligned to obtain a baseline graph; according to the timestamp information in the temporal relation set, a time interval attribute is added to each relation edge in the baseline graph, wherein the time interval attribute is jointly defined by the start timestamp and end timestamp of the relation; and according to a predetermined... The confidence level of the meaning or the statistical characteristics based on co-occurrence frequency are used to calculate the relation strength weight for each relation edge. The benchmark graph carrying the relation strength weight and time interval attribute is determined as the main graph structure. The event feature vector is generated from the entity nodes in the main graph structure using the event-entity association matrix. The event feature vector is used as the initial feature of the node by the temporal graph embedding algorithm. At the same time, the topological connection, relation strength and time interval in the main graph structure are learned to obtain the graph embedding representation. The benchmark graph and the graph embedding representation obtained by the temporal graph embedding algorithm are stored together in the graph database to obtain the dynamic knowledge graph. The graph embedding representation is used as the vectorized attribute of the graph node and is stored in association with the topological structure of the benchmark graph.

[0044] In this embodiment, an initial graph structure carrying time stamps is first constructed based on the subject, object, relationship, and timestamp information in the time-series relationship set. The timestamp information is used to accurately identify the start time and effective time period corresponding to each relationship, ensuring that the graph can reflect the timeliness characteristics of the financial entity relationship. Nodes in the initial graph are then aligned using a method based on entity attribute similarity to eliminate ambiguity between nodes pointing to the same entity from different data sources, forming a baseline graph with unified entity identifiers.

[0045] Secondly, based on the timestamp information in the time-series relation set, a time interval attribute defined by the start and end timestamps is added to each relation edge in the baseline graph. At the same time, a quantified relation strength weight is calculated for each relation edge based on the predefined confidence level or statistical characteristics based on co-occurrence frequency, so that the graph can accurately reflect both the validity period of the relation and the reliability of the relation.

[0046] The baseline graph, carrying relation strength weights and time interval attributes, is again used as the main graph structure. Event feature vectors are generated from entity nodes in the main graph structure using the event-entity association matrix. These event feature vectors capture the association patterns between entities and various financial events. A temporal graph embedding algorithm is employed, using the event feature vectors as initial node features. Simultaneously, the algorithm learns the topological connectivity characteristics, relation strength weights, and time interval information within the main graph structure. By comprehensively considering network structure, relation attributes, and temporal features, a high-quality knowledge graph embedding representation is generated.

[0047] Finally, the baseline graph and the graph embedding representation obtained through the temporal graph embedding algorithm are jointly stored in the graph database to construct a complete dynamic knowledge graph. In the dynamic knowledge graph, the graph embedding representation, as a vectorized attribute of the graph node, is associated with the topological structure of the baseline graph and stored accordingly. This preserves the explicit semantic relationships of the graph while also possessing the implicit representation capabilities of deep learning.

[0048] The above steps, by transforming timestamp information into explicit time interval attributes, overcome the static limitations of traditional knowledge graphs. This not only records the existence of relationships between financial entities but also precisely depicts the lifecycle and effective periods of these relationships. The time-series annotation mechanism enables the graph to dynamically reflect the changing relationship networks in the financial market as policies adjust and market cycles evolve, providing a data foundation for analyzing time-varying correlations of economic variables.

[0049] Entity alignment addresses the issue of data heterogeneity, confidence assessment is incorporated through relation strength weights, and external event context is introduced via event feature vectors. Finally, graph embedding learning unifies topology, relation strength, time dimension, and event features into a single vector space. This multi-layered fusion effectively overcomes the limitations of single data sources, creating a comprehensive characterization of financial entities and their relationships, significantly improving the completeness and accuracy of knowledge representation.

[0050] An innovative dual storage strategy of "benchmark graph + embedded representation" is adopted. The benchmark graph preserves the semantic transparency and interpretability of the original relationships, facilitating human understanding and verification; the graph embedding representation extracts latent semantic features through deep learning, supporting efficient similarity calculation and relationship reasoning. This design not only meets the stringent requirements of decision interpretability in the financial field but also provides computational feasibility for large-scale complex analysis.

[0051] According to some alternative embodiments of this application, the hybrid causal discovery method for identifying the causal graph structure corresponding to financial variables in a query content within a dynamic knowledge graph can be achieved through the following steps: Extracting a set of core financial variables related to the query intent of the query content from the dynamic knowledge graph, and obtaining the graph embedding representation corresponding to the core financial variables; employing a constraint-based causal discovery algorithm, using the set of core financial variables as nodes and the time priority constraints included in the time sequence relationship set in the dynamic knowledge graph as necessary conditions for determining the causal direction, constructing an initial causal skeleton graph, wherein the initial causal skeleton graph is used to characterize whether there is a causal relationship between variables; in the initial causal skeleton... Based on the graph, a score-based causal discovery algorithm is adopted. This algorithm incorporates the similarity between graph embedding representations into the graph structure scoring function and uses key event association edges in the event-entity association matrix that exceed a preset threshold as directional constraints to strengthen the search, thereby optimizing and obtaining candidate causal graph structures. Based on the timestamp information in the temporal relationship set, causal edges in the candidate causal graph structures that do not conform to the preset temporal logic are removed, and the confidence scores of the retained causal edges are calculated using historical data in the dynamic knowledge graph. Based on the temporally verified causal edges and their confidence scores, a causal graph structure is generated, which includes temporal lag attributes and confidence weights.

[0052] In this embodiment, a set of core financial variables closely related to the query intent is first extracted from the dynamic knowledge graph. These core financial variables can be selected from the graph nodes through semantic matching and relevance analysis. Simultaneously, the graph embedding representations corresponding to these variables are obtained, where the embedding vectors capture the structural features and semantic information of the variables in the graph. A constraint-based causal discovery algorithm is employed, using the set of core financial variables as the analysis nodes. It emphasizes the introduction of time priority constraints inherent in the temporal relationship set as a necessary condition for determining the causal direction; that is, the causal variable must occur before the outcome variable in time. An initial causal skeleton graph is constructed through conditional independence testing and causal edge orientation. This skeleton graph can clearly characterize whether there are potential causal relationships between variables.

[0053] Further structural optimization is achieved using a score-based causal discovery algorithm. This algorithm incorporates the similarity between graph embedding representations into the graph structure scoring function, making it more likely that semantically similar variables in the vector space will establish causal connections. Simultaneously, it uses key event association edges exceeding a preset threshold in the event-entity association matrix as directional constraints to strengthen the search, guiding the algorithm to prioritize causal directions with strong event support. By continuously optimizing the overall graph structure score, a more accurate candidate causal graph structure is obtained.

[0054] Then, based on the detailed timestamp information in the temporal relation set, a rigorous temporal logic verification is performed on the candidate causal graph structure, eliminating edges whose temporal order does not conform to causality, such as invalid connections where the result precedes the cause. Simultaneously, leveraging the rich historical data in the dynamic knowledge graph, confidence scores for retained causal edges are calculated through statistical tests and hypothesis testing to ensure that each causal edge has sufficient data support.

[0055] Finally, all causal edges that have passed time-series verification and their corresponding confidence scores are integrated to generate a complete causal graph structure that includes time-series lag attributes and confidence weights. The complete causal graph structure can not only clearly show the causal relationship network between financial variables, but also provide a measure of the strength and time characteristics of each causal relationship.

[0056] The above steps, by combining constraint-based and fraction-based causal discovery algorithms, fully utilize the statistical regularities in the data while incorporating temporal logic and prior knowledge of events from the financial field. This hybrid approach effectively overcomes the potential for spurious causal relationships arising from purely data-driven methods and avoids subjective biases inherent in purely domain-based knowledge, thus forming a more reliable and comprehensive causal discovery framework.

[0057] Innovatively, the similarity between graph embedding representations is incorporated into the graph structure scoring function, making variables with similar semantics more likely to establish causal connections; at the same time, the key event association edges in the event-entity association matrix are used as directional constraints, which can significantly improve the accuracy of causal direction judgment.

[0058] By rigorously verifying temporal logic to eliminate spurious causal relationships that violate temporal patterns, and combining this with historical data to calculate confidence scores, the resulting causal graph structure not only includes the causal relationships between variables but also provides the temporal lag characteristics and reliability measures for each causal relationship. This rich metadata transforms the causal graph from a simple connectivity graph into a complex causal network containing multi-dimensional attributes such as strength, temporality, and reliability.

[0059] In some optional embodiments of this application, a dual machine learning model is used to determine the causal effect parameters in the causal graph structure. This can be achieved through the following method: Based on the causal path in the causal graph structure, an input feature set for causal inference is determined, wherein the input feature set includes at least: a treatment variable, an outcome variable, and a set of confounding variables. The treatment variable represents the intervention to be implemented, the outcome variable represents the target indicator to be evaluated, and the set of confounding variables controls for background factors that have a common impact on the treatment variable and the outcome variable. Using a first machine learning model, with the set of confounding variables as input, the treatment variable is predicted to obtain a first predicted value of the treatment variable, and a first residual between the actual value of the treatment variable and the first predicted value is calculated. Using a second machine learning model, with the set of confounding variables as input, the outcome variable is predicted. The process involves obtaining a second predicted value for the outcome variable and calculating a second residual between the actual value and the second predicted value. The second machine learning model and the first machine learning model are independent of each other in terms of model structure or model parameters. The correlation between the first and second residuals is calculated using orthogonalization techniques to obtain the average treatment effect parameter, which characterizes the global average causal impact of the treatment variable on the outcome variable. Based on a heterogeneity analysis model, using the confounding variable set as conditional variables, the conditional average treatment effect parameter is calculated for different subgroups. The heterogeneity analysis model is a decision tree-based or meta-learner model, and the different subgroups are sample sets with similar characteristics partitioned according to the feature space of the confounding variables. The average treatment effect parameter and the conditional average treatment effect parameter are jointly determined as the causal effect parameter.

[0060] In this embodiment, the input feature set required for causal inference is first determined based on the causal path identified in the causal graph structure. The input feature set includes: processing variables representing the intervention measures to be implemented, such as interest rate adjustments or changes in credit policies; outcome variables corresponding to the target indicators to be evaluated, such as corporate financing costs or market volatility; and a set of confounding variables covering background factors that affect both processing and outcome variables, such as the macroeconomic environment and industry cycles. By controlling these confounding factors, the accuracy of causal inference can be ensured.

[0061] Then, two independent machine learning models are established for parameter estimation. Using the first machine learning model, with the set of confounding variables as input features, the treatment variable is predicted to obtain a first predicted value. The first residual between the actual and predicted values ​​of the treatment variable is then calculated; this first residual represents the net change in the treatment variable after removing the confounding factors. Simultaneously, using the second machine learning model, also with the set of confounding variables as input, the outcome variable is predicted to obtain a second predicted value. The second residual between the actual and predicted values ​​of the outcome variable is then calculated; this second residual reflects the independent variation of the outcome variable after removing the confounding factors. It should be noted that the two machine learning models are kept independent in terms of model structure and parameters, which effectively avoids overfitting and ensures the robustness of the estimation.

[0062] Secondly, orthogonalization techniques were used to analyze the intrinsic relationship between the first and second residuals, and the average treatment effect parameter was calculated. The average treatment effect parameter characterizes the global average causal impact of the treatment variable on the outcome variable. Its calculation process effectively eliminates biases introduced by confounding variables, providing a reliable estimate of the overall effect of the intervention. To further reveal the heterogeneity of causal effects, based on heterogeneity analysis models such as decision trees or meta-learners, the conditional average treatment effect parameter was calculated under different characteristic subgroups, using the set of confounding variables as conditional variables. Here, the subgroup is a set of samples with similar characteristics, partitioned by cluster analysis based on the characteristic space of the confounding variables, enabling the causal effect estimation to reflect the differentiated impacts under different market environments and enterprise characteristics.

[0063] Finally, the average treatment effect parameter and the conditional average treatment effect parameter were jointly determined as the causal effect parameter. These two parameters comprehensively characterize the causal influence of the treatment variable on the outcome variable from two dimensions: overall average and group heterogeneity, respectively. This dual machine learning framework, combined with heterogeneity analysis, effectively overcomes the sensitivity of administrative causal inference methods to model specification, and can provide more accurate and detailed causal effect assessments in complex financial scenarios.

[0064] The above steps, by employing a dual machine learning model, effectively address the estimation bias problem caused by confounding variables in traditional causal inference. Two independent machine learning models predict the treatment variable and the outcome variable respectively, and the influence of confounding factors is eliminated through residual calculation. This approach can obtain the pure causal effect of the treatment variable on the outcome variable without relying on strong model assumptions, significantly improving the accuracy and reliability of the estimation results.

[0065] Furthermore, this study not only provides a global assessment of the impact of interventions through the average treatment effect parameter, but also reveals the heterogeneous manifestation of causal effects across different subgroups through the conditional average treatment effect parameter. This dual quantification mechanism enables policymakers to grasp both overall trends and identify differentiated impacts, providing fine-grained data support for precise policy implementation.

[0066] As some optional embodiments of this application, a structural causal model can be constructed based on the causal graph structure and causal effect parameters, which can be achieved by the following methods: A structural equation framework is constructed based on the nodes and edges in the causal graph structure, wherein each structural equation in the structural equation framework adopts a linear function form, representing the current variable as a linear combination of all parent variables in the causal graph structure; based on the causal effect parameters, the initial estimated values ​​of each path coefficient in the structural equation framework are inversely estimated by calculating the path coefficients based on the structural equation model, thus obtaining the initial structural equation; based on historical data in the dynamic knowledge graph, the maximum likelihood estimation method is used to perform joint parameter estimation on the initial structural equation, resulting in an optimized structural equation parameter set, wherein the optimized structural equation parameter set includes: point estimates of path coefficients and confidence intervals of path coefficients; the optimized structural equation parameter set is then configured onto the corresponding edges of the causal graph structure to obtain the structural causal model.

[0067] In this embodiment, a structural equation framework is constructed based on the nodes and directed edge relationships determined in the causal graph structure. Each structural equation in the framework adopts a linear function form, representing the current variable as a linear combination of all its parent variables in the causal graph structure, thereby establishing the mathematical basis for the causal dependencies between variables. Based on the estimated causal effect parameters, the initial estimated values ​​of each path coefficient in the structural equation framework are derived by using the path coefficient calculation method in the structural equation model, thus completing the establishment of the initial structural equations.

[0068] Leveraging the rich historical data within a dynamic knowledge graph, the maximum likelihood estimation method is employed to perform joint parameter estimation on the initial structural equation model. The optimization process adjusts the path coefficients by maximizing the likelihood function of the observed data, fully considering the collaborative change patterns among variables and the statistical characteristics of the data, resulting in an optimized set of structural equation parameters. This parameter set not only contains precise point estimates of each path coefficient but also provides corresponding confidence intervals, thus comprehensively characterizing the accuracy and reliability of the parameter estimation.

[0069] The optimized set of structural equation parameters is configured onto the corresponding edges of the causal graph structure, thus completing the construction of the structural causal model. This configuration process transforms the original causal graph structure into a structural causal model with explicit mathematical relationships and quantified parameters. Each causal edge is assigned specific path coefficients and confidence intervals, forming a complete causal representation that is both interpretable and computable.

[0070] The above steps successfully transformed qualitative causal relationships into quantitative mathematical expressions by converting the causal graph structure into a computable structural equation framework. Each structural equation precisely corresponds to a causal path in the causal graph, giving the abstract causal relationship a concrete functional form. A phased parameter estimation strategy is adopted: first, the initial values ​​of the path coefficients are inferred from the parameters of previous causal effects; then, joint optimization is performed using maximum likelihood estimation based on historical data. This hybrid approach of "prior guidance + data-driven" fully utilizes the results of previous causal discoveries and verifies and corrects parameter estimates through historical data, significantly improving the accuracy and robustness of parameter estimation and effectively avoiding the local optima problem caused by improper parameter initialization in traditional methods.

[0071] In some optional embodiments of this application, generating counterfactual prediction results using a structural causal model can be achieved through the following methods: determining a set of intervention operations based on query content, wherein each intervention operation in the set is represented by a preset operator, which is used to simulate intervention on preset processing variables in the causal graph structure; for each intervention operation in the set, calculating the potential results of the intervention operation on the outcome variable based on the optimized structural equation in the structural causal model to obtain a basic counterfactual prediction; quantifying the uncertainty of the basic counterfactual prediction based on the confidence intervals of the path coefficients in the optimized structural equation parameter set to obtain a counterfactual prediction distribution including the confidence intervals; and integrating the counterfactual prediction distributions corresponding to all intervention operations in the set to obtain the counterfactual prediction result.

[0072] In this embodiment, the set of intervention operations to be simulated is first determined based on the explicit business requirements in the query content. Each intervention operation is formally represented by a pre-defined do-operator, which can be used to simulate the semantics of applying external intervention to specific processing variables in the causal graph structure, ensuring the logical rigor of the intervention. For each specific intervention in the set of intervention operations, calculations are performed based on the optimized structural equation in the structural causal model. After setting the intervention variable to a specific value using the substitution method, the influence propagates along the causal path, accurately calculating the potential result of the intervention operation on the target outcome variable, thereby obtaining the basic counterfactual prediction value. Further, uncertainty quantification analysis is performed based on the confidence intervals of each path coefficient in the optimized structural equation parameter set. Specifically, the Monte Carlo simulation method can be used, which simulates the transmission of parameter uncertainty to the prediction result by repeatedly sampling from the posterior distribution of the parameters, obtaining a counterfactual prediction distribution containing confidence intervals. The counterfactual prediction distribution not only provides the central trend of the prediction, but more importantly, it characterizes the range of uncertainty in the prediction. Finally, the counterfactual prediction distributions corresponding to all intervention operations in the intervention operation set are integrated to form a complete counterfactual prediction result. The counterfactual prediction result presents the effect comparison under different intervention scenarios in a standardized data structure, including the expected effect value, confidence interval and distribution characteristics of each intervention.

[0073] As some alternative embodiments of this application, generating a decision scheme corresponding to the query content based on the counterfactual prediction results can be achieved through the following methods: Based on the business objectives of the query content, determine the optimization objectives and a set of constraints, wherein the optimization objectives include at least: maximizing revenue and minimizing risk, and the set of constraints comes from business rules and risk limits in the dynamic knowledge graph; establish a multi-objective optimization function based on the expected effect value in the counterfactual prediction results as the basis for revenue estimation and the confidence interval in the counterfactual prediction results as the basis for risk assessment; construct risk constraints based on the risk transmission paths identified in the causal graph structure and add the risk constraints to the set of constraints; solve the multi-objective optimization function using a multi-objective optimization algorithm to obtain the optimal solution set, wherein each solution in the optimal solution set corresponds to a candidate decision strategy and the expected performance of the candidate decision strategy on each optimization objective; calculate the robustness score of each strategy based on the causal path length and node influence range of the candidate decision strategy in the causal graph structure; select the optimal decision strategy from the optimal solution set according to the robustness score, and transform the optimal decision strategy into an executable decision scheme.

[0074] In this embodiment, the objective function and constraints of the optimization problem are first determined based on the explicit business objectives in the query content. The optimization objective must at least cover two core dimensions: maximizing revenue and minimizing risk. The set of constraints is extracted from the dynamic knowledge graph, including relevant business rules and risk limit requirements, to ensure that the decision-making scheme complies with business specifications and risk control standards. The expected effect value in the counterfactual prediction results is used as the quantitative basis for the revenue objective. Simultaneously, the confidence interval of the prediction results is used to construct a risk assessment index. By integrating these two elements into a mathematical programming model to establish a multi-objective optimization function, a comprehensive optimization framework that considers both expected revenue and uncertainty can be formed.

[0075] To further improve the risk control mechanism, specialized risk constraints are constructed based on the risk transmission paths identified in the causal graph structure. By analyzing the propagation paths and impact range of risk factors in the causal network, risk considerations are transformed into mathematical constraint expressions, and these risk constraints are added to the original set of constraints, enabling the optimization model to effectively avoid undesirable decisions that may trigger cascading risks. A multi-objective optimization algorithm is used to solve the established optimization function, and the optimal solution set is obtained through Pareto optimality front analysis. Each solution in the optimal solution set corresponds to a candidate decision strategy and its expected performance on various optimization objectives, providing a diverse decision choice space.

[0076] Furthermore, a robustness score is calculated for each candidate decision strategy based on its characteristics within the causal graph structure. By analyzing the causal path length and node influence range corresponding to each strategy, the sensitivity of the strategy's execution effect to model assumptions and parameter estimates is evaluated. Strategies with shorter paths and concentrated influence ranges generally exhibit higher robustness. Finally, based on the calculated robustness scores, the best-performing decision strategy is selected from the optimal solution set. This optimal strategy is then transformed into an executable decision plan that includes specific operational steps, expected effect assessments, and risk control measures, ensuring that the final output decision not only meets the optimization objective requirements but also possesses good implementation stability.

[0077] The above steps successfully construct a complete link from causal reasoning to decision generation by directly using counterfactual prediction results as input to the optimization model. Unlike traditional optimization methods that rely on historical statistical patterns, this approach predicts the potential effects of intervention measures based on causal models. This enables the optimization process to accurately predict the chain reactions of various decisions in complex causal networks, thereby generating more forward-looking decision solutions.

[0078] Figure 2 This is a structural diagram of a management decision-making system based on knowledge base construction technology according to an embodiment of this application, such as... Figure 2 As shown, the system includes:

[0079] The receiving module 21 is used to receive query content and extract time-series relationships from multi-source financial data based on the query content to obtain a set of time-series relationships related to the query content. The set of time-series relationships is a structured data set used to dynamically represent the relationship between financial entities.

[0080] The first construction module 22 is used to construct the event-entity association matrix corresponding to the time-series relationship set, wherein the event-entity association matrix is ​​used to quantify the association strength between financial events and financial entities.

[0081] The second construction module 23 is used to construct a dynamic knowledge graph based on the time sequence relationship set and the event-entity association matrix, and to identify the causal graph structure corresponding to the financial variables in the query content in the dynamic knowledge graph using a hybrid causal discovery method. The hybrid causal discovery method includes a constraint-based causal discovery algorithm and a score-based causal discovery algorithm. The causal graph structure is a graphical representation of the causal relationship and transmission path between financial variables.

[0082] Module 24 is used to determine the causal effect parameters in the causal graph structure using a dual machine learning model. The causal effect parameters are used to quantify the degree of causal influence of interventions targeting the query content on the target variable.

[0083] The third building module 25 is used to construct a structural causal model based on the causal graph structure and causal effect parameters, wherein the structural causal model includes structural equations for defining the causal mechanism between financial variables.

[0084] The generation module 26 is used to generate counterfactual prediction results using a structural causal model, and to generate decision schemes corresponding to the query content based on the counterfactual prediction results.

[0085] Optionally, a dynamic knowledge graph is constructed based on the temporal relation set and the event-entity association matrix, including: constructing an initial graph structure with time stamps based on the subject, object, relation, and timestamp information in the temporal relation set, wherein the timestamp information is used to identify the start time or effective time period of the relation; aligning the nodes in the initial graph structure based on the similarity of entity attributes to obtain a baseline graph; adding a time interval attribute to each relation edge in the baseline graph based on the timestamp information in the temporal relation set, wherein the time interval attribute is jointly defined by the start and end timestamps of the relation; and based on the predefined confidence level in the temporal relation set or based on the common... The statistical characteristics of current frequency are used to calculate the relationship strength weight for each relation edge. The baseline graph carrying the relationship strength weight and time interval attribute is determined as the main graph structure. Event feature vectors are generated from the entity nodes in the main graph structure using the event-entity association matrix. The event feature vectors are used as the initial features of the nodes using a time-series graph embedding algorithm. At the same time, the topological connections, relationship strength, and time interval in the main graph structure are learned to obtain the graph embedding representation. The baseline graph and the graph embedding representation obtained by the time-series graph embedding algorithm are stored together in the graph database to obtain a dynamic knowledge graph. The graph embedding representation is used as a vectorized attribute of the graph node and is stored in association with the topological structure of the baseline graph.

[0086] Optionally, a hybrid causal discovery method is employed to identify the causal graph structure corresponding to financial variables in the query content within a dynamic knowledge graph. This includes: extracting a set of core financial variables related to the query intent of the query content from the dynamic knowledge graph and obtaining the graph embedding representation corresponding to the core financial variables; constructing an initial causal skeleton graph using a constraint-based causal discovery algorithm, with the set of core financial variables as nodes and the time priority constraints included in the time sequence relation set in the dynamic knowledge graph as necessary conditions for determining the causal direction, wherein the initial causal skeleton graph is used to characterize whether there is a causal relationship between variables; and, based on the initial causal skeleton graph, employing a constraint-based causal discovery algorithm... The causal discovery algorithm for scores incorporates the similarity between graph embedding representations into the graph structure scoring function and uses key event association edges in the event-entity association matrix that exceed a preset threshold as directional constraints to strengthen the search, thereby optimizing the candidate causal graph structure. Based on the timestamp information in the temporal relationship set, causal edges in the candidate causal graph structure that do not conform to the preset temporal logic are removed, and the confidence score of the retained causal edges is calculated using historical data in the dynamic knowledge graph. Based on the temporally verified causal edges and their confidence scores, a causal graph structure is generated, which includes temporal lag attributes and confidence weights.

[0087] Optionally, a dual machine learning model is employed to determine the causal effect parameters in the causal graph structure, including: determining the input feature set for causal inference based on the causal path in the causal graph structure, wherein the input feature set includes at least: a treatment variable, an outcome variable, and a set of confounding variables, where the treatment variable represents the intervention to be implemented, the outcome variable represents the target indicator to be evaluated, and the set of confounding variables controls for background factors that jointly affect the treatment variable and the outcome variable; using a first machine learning model, with the set of confounding variables as input, predicting the treatment variable to obtain a first predicted value of the treatment variable and calculating a first residual between the actual value of the treatment variable and the first predicted value; using a second machine learning model, with the set of confounding variables as input, predicting the outcome variable to obtain a first residual between the actual value of the treatment variable and the first predicted value. The second predicted value is used to calculate the second residual between the actual value of the outcome variable and the second predicted value. The second machine learning model and the first machine learning model are independent of each other in terms of model structure or model parameters. The correlation between the first residual and the second residual is calculated using orthogonalization techniques to obtain the average treatment effect parameter. The average treatment effect parameter is used to characterize the global average causal effect of the treatment variable on the outcome variable. Based on the heterogeneity analysis model, the conditional average treatment effect parameter is calculated for different subgroups using the confounding variable set as the conditional variable. The heterogeneity analysis model is a model based on decision trees or meta-learners, and the different subgroups are sample sets with similar characteristics divided according to the feature space of the confounding variables. The average treatment effect parameter and the conditional average treatment effect parameter are jointly determined as the causal effect parameter.

[0088] Optionally, a structural causal model is constructed based on the causal graph structure and causal effect parameters, including: constructing a structural equation framework based on the nodes and edges in the causal graph structure, wherein each structural equation in the structural equation framework adopts a linear function form, representing the current variable as a linear combination of all parent variables of the current variable in the causal graph structure; based on the causal effect parameters, the initial estimated values ​​of each path coefficient in the structural equation framework are inversely estimated by calculating the path coefficients based on the structural equation model, thereby obtaining the initial structural equation; based on historical data in the dynamic knowledge graph, the maximum likelihood estimation method is used to perform joint parameter estimation on the initial structural equation, thereby obtaining an optimized structural equation parameter set, wherein the optimized structural equation parameter set includes: point estimates of path coefficients and confidence intervals of path coefficients; and configuring the optimized structural equation parameter set onto the corresponding edges of the causal graph structure to obtain the structural causal model.

[0089] Optionally, generating counterfactual prediction results using a structural causal model includes: determining a set of intervention operations based on query content, wherein each intervention operation in the set is represented by a preset operator, which is used to simulate intervention on preset processing variables in the causal graph structure; for each intervention operation in the set, calculating the potential results of the intervention operation on the outcome variable based on the optimized structural equation in the structural causal model to obtain a basic counterfactual prediction; quantifying the uncertainty of the basic counterfactual prediction based on the confidence intervals of the path coefficients in the optimized structural equation parameter set to obtain a counterfactual prediction distribution including the confidence intervals; and integrating the counterfactual prediction distributions corresponding to all intervention operations in the set to obtain the counterfactual prediction result.

[0090] Optionally, based on the counterfactual prediction results, a decision scheme corresponding to the query content is generated, including: determining the optimization objective and constraint set based on the business objectives of the query content, wherein the optimization objective includes at least: maximizing revenue and minimizing risk, and the constraint set comes from business rules and risk limits in the dynamic knowledge graph; establishing a multi-objective optimization function based on the expected effect value in the counterfactual prediction results as the basis for revenue estimation and the confidence interval in the counterfactual prediction results as the basis for risk assessment; constructing risk constraints based on the risk transmission paths identified in the causal graph structure and adding the risk constraints to the constraint set; solving the multi-objective optimization function using a multi-objective optimization algorithm to obtain the optimal solution set, wherein each solution in the optimal solution set corresponds to a candidate decision strategy and the expected performance of the candidate decision strategy on each optimization objective; calculating the robustness score of each strategy based on the causal path length and node influence range of the candidate decision strategy in the causal graph structure; selecting the optimal decision strategy from the optimal solution set according to the robustness score, and transforming the optimal decision strategy into an executable decision scheme.

[0091] It should be noted that the above Figure 2 The modules in can be program modules (e.g., a set of program instructions that implements a specific function) or hardware modules. For the latter, they can be represented in the following forms, but are not limited to these: each of the above modules is represented by a processor, or the functions of each of the above modules are implemented by a processor.

[0092] It should be noted that, Figure 2 Preferred embodiments of the shown examples can be found in [reference needed]. Figure 1 The relevant descriptions of the embodiments shown will not be repeated here.

[0093] Figure 3 A hardware block diagram of a computer terminal for implementing a management decision-making method based on knowledge base construction technology is shown. Figure 3As shown, the computer terminal 30 may include one or more processors 302 (shown as 302a, 302b, ..., 302n in the figure) 302 (processor 302 may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 304 for storing data, and a transmission module 306 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 3 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 30 may also include... Figure 3 The more or fewer components shown, or having the same Figure 3 The different configurations shown.

[0094] It should be noted that the aforementioned one or more processors 302 and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element within the computer terminal 30. As involved in the embodiments of this application, the data processing circuits serve as processor control (e.g., selection of a variable resistor termination path connected to an interface).

[0095] The memory 304 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the management decision-making method based on knowledge base construction technology in this embodiment of the application. The processor 302 executes various functional applications and data processing by running the software programs and modules stored in the memory 304, thereby realizing the aforementioned management decision-making method based on knowledge base construction technology. The memory 304 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 304 may further include memory remotely located relative to the processor 302, and these remote memories can be connected to the computer terminal 30 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0096] The transmission module 306 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the computer terminal 30. In one example, the transmission module 306 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission module 306 may be a Radio Frequency (RF) module, used for wireless communication with the Internet.

[0097] The display may be, for example, a touchscreen liquid crystal display (LCD) that allows the user to interact with the user interface of the computer terminal 30.

[0098] It should be noted here that, in some optional embodiments, the above... Figure 3 The computer terminal shown may include hardware elements (including circuitry), software elements (including computer code stored on a computer-readable medium), or a combination of both hardware and software elements. It should be noted that... Figure 3 This is only one instance of a specific particular instance, and is intended to illustrate the types of components that may exist in the aforementioned computer terminal.

[0099] It should be noted that, Figure 3 The computer terminal shown is used to execute Figure 1 The management decision-making method based on knowledge base construction technology shown above also applies to this electronic device, and will not be repeated here.

[0100] This application also provides a non-volatile storage medium, which includes a stored program, wherein the program, when running, controls the device where the storage medium is located to execute the above-mentioned management decision-making method based on knowledge base construction technology.

[0101] A program on a non-volatile storage medium performs the following functions: receiving query content; extracting time-series relationships from multi-source financial data based on the query content to obtain a set of time-series relationships related to the query content, wherein the set of time-series relationships is a structured data set used to dynamically represent the relationships between financial entities; constructing an event-entity association matrix corresponding to the set of time-series relationships, wherein the event-entity association matrix is ​​used to quantify the association strength between financial events and financial entities; constructing a dynamic knowledge graph based on the set of time-series relationships and the event-entity association matrix, and using a hybrid causal discovery method to identify the causal graph structure corresponding to the financial variables in the query content in the dynamic knowledge graph, wherein the hybrid causal discovery method... The causal discovery method includes: a constraint-based causal discovery algorithm and a score-based causal discovery algorithm. The causal graph structure is a graphical representation of the causal relationships and transmission paths between financial variables. A dual machine learning model is used to determine the causal effect parameters in the causal graph structure, whereby the causal effect parameters are used to quantitatively assess the causal impact of intervention measures targeting the query content on the target variable. Based on the causal graph structure and causal effect parameters, a structural causal model is constructed, whereby the structural causal model includes structural equations for defining the causal mechanisms between financial variables. Counterfactual prediction results are generated using the structural causal model, and based on the counterfactual prediction results, decision schemes corresponding to the query content are generated.

[0102] This application also provides an electronic device, including a memory and a processor, wherein the processor is used to run a program stored in the memory, wherein the program executes the above-described management decision-making method based on knowledge base construction technology.

[0103] The processor runs a program that performs the following functions: receiving query content; extracting time-series relationships from multi-source financial data based on the query content to obtain a set of time-series relationships related to the query content, wherein the set of time-series relationships is a structured data set used to dynamically represent the relationships between financial entities; constructing an event-entity association matrix corresponding to the set of time-series relationships, wherein the event-entity association matrix is ​​used to quantify the association strength between financial events and financial entities; constructing a dynamic knowledge graph based on the set of time-series relationships and the event-entity association matrix, and using a hybrid causal discovery method to identify the causal graph structure corresponding to the financial variables in the query content in the dynamic knowledge graph, wherein the hybrid... The causal discovery methods include: constraint-based causal discovery algorithms and score-based causal discovery algorithms. The causal graph structure is a graphical representation of the causal relationships and transmission paths between financial variables. A dual machine learning model is employed to determine the causal effect parameters in the causal graph structure, whereby these parameters are used to quantitatively assess the causal impact of interventions targeting the query content on the target variable. Based on the causal graph structure and causal effect parameters, a structural causal model is constructed, which includes structural equations defining the causal mechanisms between financial variables. The structural causal model is used to generate counterfactual prediction results, and based on these results, decision-making schemes corresponding to the query content are generated.

[0104] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0105] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0106] In the above embodiments of this application, the information collected is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data all comply with relevant laws, regulations and standards, take necessary protective measures, do not violate public order and good morals, and provide corresponding operation entry points for users to choose to authorize or refuse.

[0107] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0108] 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 units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0109] Furthermore, the functional units in the various embodiments of this application 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. The integrated unit can be implemented in hardware or as a software functional unit.

[0110] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to related technologies, or all or 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 application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0111] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A management decision-making method based on knowledge base construction technology, characterized in that, include: Receive query content, and extract time-series relationships from multi-source financial data based on the query content to obtain a time-series relationship set related to the query content. The time-series relationship set is a structured data set used to dynamically represent the relationship between financial entities. Construct an event-entity association matrix corresponding to the time-series relationship set, wherein the event-entity association matrix is ​​used to quantitatively represent the association strength between financial events and financial entities; Based on the time-series relationship set and the event-entity association matrix, a dynamic knowledge graph is constructed, and a hybrid causal discovery method is used to identify the causal graph structure corresponding to the financial variables in the query content in the dynamic knowledge graph. The hybrid causal discovery method includes a constraint-based causal discovery algorithm and a score-based causal discovery algorithm. The causal graph structure is a graphical representation of the causal relationships and transmission paths between the financial variables. A dual machine learning model is used to determine the causal effect parameters in the causal graph structure, wherein the causal effect parameters are used to quantitatively evaluate the degree of causal influence of intervention measures on the query content on the target variable; Based on the causal graph structure and the causal effect parameters, a structural causal model is constructed, wherein the structural causal model includes structural equations for defining the causal mechanism between the financial variables; The structural causal model is used to generate counterfactual prediction results, and based on the counterfactual prediction results, a decision scheme corresponding to the query content is generated.

2. The method according to claim 1, characterized in that, Based on the aforementioned temporal relationship set and event-entity association matrix, a dynamic knowledge graph is constructed, including: Based on the subject, object, relationship, and timestamp information in the time-series relationship set, an initial graph structure carrying time stamps is constructed, wherein the timestamp information is used to identify the start time or effective time period of the relationship; Based on the similarity of entity attributes, the nodes in the initial graph structure are aligned to obtain a baseline graph. Based on the timestamp information in the time-series relation set, a time interval attribute is added to each relation edge in the baseline graph, wherein the time interval attribute is defined by the start and end timestamps of the relation; and a relation strength weight is calculated for each relation edge based on the predefined confidence level or statistical features based on co-occurrence frequency in the time-series relation set. The baseline graph carrying the relation strength weights and the time interval attributes is determined as the main graph structure. The event-entity association matrix is ​​used to generate event feature vectors for entity nodes in the main graph structure. The time-series graph embedding algorithm is used to take the event feature vectors as the initial features of the nodes. At the same time, the topological connections, relation strengths and time intervals in the main graph structure are learned to obtain the graph embedding representation. The baseline graph and the graph embedding representation obtained by the temporal graph embedding algorithm are jointly stored in the graph database to obtain the dynamic knowledge graph. The graph embedding representation is stored as a vectorized attribute of the graph node and is associated with the topological structure of the baseline graph.

3. The method according to claim 1, characterized in that, A hybrid causal discovery method is used to identify the causal graph structure corresponding to the financial variables in the query content within the dynamic knowledge graph, including: Extract a set of core financial variables related to the query intent of the query content from the dynamic knowledge graph, and obtain the graph embedding representation corresponding to the core financial variables; A constraint-based causal discovery algorithm is adopted, with the core financial variable set as nodes and the time priority constraints included in the time sequence relationship set in the dynamic knowledge graph as necessary conditions for determining the causal direction, to construct an initial causal skeleton graph, wherein the initial causal skeleton graph is used to characterize whether there is a causal relationship between variables; Based on the initial causal skeleton graph, a score-based causal discovery algorithm is adopted. By incorporating the similarity between the graph embedding representations into the graph structure scoring function, and using the key event association edges in the event-entity association matrix that are higher than a preset threshold as directional constraints to strengthen the search, the candidate causal graph structure is optimized. Based on the timestamp information in the time sequence relation set, causal edges that do not conform to the preset time logic in the candidate causal graph structure are removed, and the confidence score of the retained causal edges is calculated using historical data in the dynamic knowledge graph. The causal graph structure is generated based on the time-tested causal edges and their confidence scores, wherein the causal graph structure includes: time-lag attributes and confidence weights.

4. The method according to claim 1, characterized in that, A dual machine learning model is used to determine the causal effect parameters in the causal graph structure, including: Based on the causal paths in the causal graph structure, an input feature set for causal inference is determined, wherein the input feature set includes at least: a treatment variable, an outcome variable, and a set of confounding variables, wherein the treatment variable is used to represent the intervention measures to be implemented, the outcome variable is used to represent the target indicators to be evaluated, and the set of confounding variables is used to control for background factors that have a common impact on the treatment variable and the outcome variable; Using a first machine learning model, with the set of confounding variables as input, the processing variable is predicted to obtain a first predicted value of the processing variable and a first residual between the actual value of the processing variable and the first predicted value is calculated. Using a second machine learning model, with the set of confounding variables as input, the result variable is predicted to obtain a second predicted value of the result variable and a second residual between the actual value of the result variable and the second predicted value is calculated. The second machine learning model and the first machine learning model are independent of each other in terms of model structure or model parameters. The correlation between the first residual and the second residual is calculated by orthogonalization technique to obtain the average treatment effect parameter, wherein the average treatment effect parameter is used to characterize the global average causal effect of the treatment variable on the outcome variable; Based on the heterogeneity analysis model, the conditional average treatment effect parameters under different subgroups are calculated using the set of confounding variables as conditional variables. The heterogeneity analysis model is a model based on decision trees or meta-learners, and the different subgroups are sample sets with similar characteristics divided according to the feature space of the confounding variables. The average treatment effect parameter and the conditional average treatment effect parameter are jointly determined as the causal effect parameter.

5. The method according to claim 4, characterized in that, Based on the causal graph structure and the causal effect parameters, a structural causal model is constructed, including: Based on the nodes and edges in the causal graph structure, a structural equation framework is constructed, wherein each structural equation in the structural equation framework adopts a linear function form, and the current variable is represented as a linear combination of all parent variables of the current variable in the causal graph structure. Based on the causal effect parameters, the initial estimates of each path coefficient in the structural equation model are inferred by calculating the path coefficients based on the structural equation model, and the initial structural equation is obtained. Based on historical data in the dynamic knowledge graph, the maximum likelihood estimation method is used to perform joint parameter estimation on the initial structural equation to obtain an optimized structural equation parameter set, wherein the optimized structural equation parameter set includes: point estimates of path coefficients and confidence intervals of the path coefficients; The optimized set of structural equation parameters is configured onto the corresponding edges of the causal graph structure to obtain the structural causal model.

6. The method according to claim 5, characterized in that, Generating counterfactual prediction results using the aforementioned structural causal model includes: An intervention operation set is determined based on the query content, wherein each intervention operation in the intervention operation set is represented by a preset operator, and the preset operator is used to simulate the intervention on a preset processing variable in the causal graph structure; For each intervention in the set of interventions, based on the optimized structural equation in the structural causal model, the potential consequences of the intervention on the outcome variable are calculated to obtain the basic counterfactual prediction; Based on the confidence intervals of the path coefficients in the optimized structural equation parameter set, the uncertainty of the basic counterfactual prediction is quantified to obtain the counterfactual prediction distribution including the confidence intervals. The counterfactual prediction distributions corresponding to all intervention operations in the intervention operation set are integrated to obtain the counterfactual prediction results.

7. The method according to claim 1, characterized in that, Based on the counterfactual prediction results, a decision scheme corresponding to the query content is generated, including: Based on the business objectives of the query content, an optimization objective and a set of constraints are determined, wherein the optimization objective includes at least: maximizing revenue and minimizing risk, and the set of constraints comes from the business rules and risk limits in the dynamic knowledge graph; Based on the expected effect value in the counterfactual prediction results, and based on the confidence interval in the counterfactual prediction results, a multi-objective optimization function is established. Based on the risk transmission path identified in the causal graph structure, risk constraints are constructed and added to the constraint set. The multi-objective optimization function is solved using a multi-objective optimization algorithm to obtain an optimal solution set, wherein each solution in the optimal solution set corresponds to a candidate decision strategy and the expected performance of the candidate decision strategy on each optimization objective; Based on the causal path length and node influence range of the candidate decision-making strategies in the causal graph structure, a robustness score is calculated for each strategy. Based on the robustness score, an optimal decision strategy is selected from the optimal solution set, and the optimal decision strategy is transformed into an executable decision scheme.

8. A management decision-making system based on knowledge base construction technology, characterized in that, include: The receiving module is used to receive query content and extract time-series relationships from multi-source financial data based on the query content to obtain a time-series relationship set related to the query content. The time-series relationship set is a structured data set used to dynamically represent the relationship between financial entities. The first construction module is used to construct the event-entity association matrix corresponding to the time-series relationship set, wherein the event-entity association matrix is ​​used to quantitatively represent the association strength between financial events and financial entities; The second construction module is used to construct a dynamic knowledge graph based on the time-series relationship set and the event-entity association matrix, and to identify the causal graph structure corresponding to the financial variables in the query content in the dynamic knowledge graph using a hybrid causal discovery method. The hybrid causal discovery method includes a constraint-based causal discovery algorithm and a score-based causal discovery algorithm. The causal graph structure is a graphical representation of the causal relationships and transmission paths between the financial variables. The determination module is used to determine the causal effect parameters in the causal graph structure using a dual machine learning model, wherein the causal effect parameters are used to quantitatively evaluate the degree of causal influence of intervention measures targeting the query content on the target variable; The third construction module is used to construct a structural causal model based on the causal graph structure and the causal effect parameters, wherein the structural causal model includes structural equations for defining the causal mechanism between the financial variables; The generation module is used to generate counterfactual prediction results using the structural causal model, and to generate a decision scheme corresponding to the query content based on the counterfactual prediction results.

9. A non-volatile storage medium, characterized in that, The non-volatile storage medium includes a stored program, wherein, when the program is executed, it controls the device where the non-volatile storage medium is located to execute the management decision-making method based on knowledge base construction technology as described in any one of claims 1 to 7.

10. An electronic device, characterized in that, include: A memory and a processor, the processor being configured to run a program stored in the memory, wherein the program, when running, executes the management decision-making method based on knowledge base construction technology as described in any one of claims 1 to 7.

Images