Processing system, method and storage medium for business data

Abstract

The application provides a business data processing system, method and storage medium. The business data processing system comprises a business graph model and a first agent. The business graph model comprises a business process sub-model, an organizational structure sub-model and a business specification sub-model. The business process sub-model, the organizational structure sub-model and the business specification sub-model each comprise a plurality of nodes and edges connecting the nodes. The first agent is configured to: obtain business data from the nodes of the business process sub-model, the organizational structure sub-model and the business specification sub-model; and modify the nodes and / or edges of at least one of the business process sub-model, the organizational structure sub-model and the business specification sub-model according to the business data. The application can be used to effectively improve the reliability, safety and certainty of the agent in performing tasks in the production field.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence and knowledge engineering technology, and in particular to a business data processing system, a business data processing method, and a computer-readable storage medium. Background Technology

[0002] Intelligent agents have rapidly become ubiquitous in everyday life, effectively meeting user needs in scenarios with clear task boundaries and ample fault tolerance, such as ordering food, hailing transportation, and information retrieval. However, in industrial settings, where reliability, security, and determinism are paramount, the probabilistic output of intelligent agents fundamentally clashes with the zero-fault-tolerance requirements of production lines. Furthermore, production systems involve complex processes such as equipment control, process parameters, and supply chain coordination, requiring strict adherence to physical laws and industry standards. Current large-scale models still suffer from the risk of illusion; misjudgments could lead to production line shutdowns, safety accidents, or substantial economic losses. Moreover, data silos and complex proprietary protocols are prevalent within enterprises, making it difficult for intelligent agents to acquire high-quality real-time data across systems. Additionally, production processes heavily rely on domain expert experience, and general-purpose models lack the accumulation of vertical industry knowledge. Furthermore, while the three mainstream technical paradigms of LEK, KEL, and LKC, which currently integrate large models and knowledge graphs, can respectively achieve automated knowledge graph construction, retrieval enhancement and generation to suppress model illusions, and bidirectional co-evolution of knowledge and models, they are only general-level technical integration paths. They cannot fully represent the complete business operation mechanism within an enterprise, and it is difficult to construct an enterprise-level business knowledge ontology that includes strict permissions, behaviors, and governance constraints.

[0003] In order to overcome the above-mentioned defects in the existing technology, there is an urgent need in the field for an improved business data processing system to effectively improve the reliability, security and determinism of intelligent agents performing tasks in the production field. Summary of the Invention

[0004] The following provides a brief overview of one or more aspects to offer a basic understanding of them. This overview is not an exhaustive summary of all conceived aspects, nor is it intended to identify key or decisive elements of all aspects, nor to define the scope of any or all aspects. Its sole purpose is to present some concepts of one or more aspects in a simplified form as a prelude to the more detailed descriptions that follow.

[0005] To overcome the aforementioned deficiencies in the existing technology, this invention provides a business data processing system, a business data processing method, and a computer-readable storage medium. By constructing a multi-level composite graph model and modifying the nodes and / or edges of multiple sub-models in the composite graph model according to the acquired business data, it can effectively improve the reliability, security, and determinism of intelligent agents performing tasks in the production field.

[0006] Specifically, according to a first aspect of the present invention, a system for processing the aforementioned business data includes a business graph model and a first intelligent agent. The business graph model includes a business process sub-model, an organizational structure sub-model, and a business specification sub-model. Each of the business process sub-model, the organizational structure sub-model, and the business specification sub-model includes multiple nodes and edges connecting each node. The first intelligent agent is configured to: acquire business data from each node of the business process sub-model, the organizational structure sub-model, and the business specification sub-model; and modify at least one of the nodes and / or edges of the business process sub-model, the organizational structure sub-model, and the business specification sub-model based on the business data.

[0007] Further, in some embodiments of the present invention, the business process sub-model includes multiple first nodes, multiple second nodes, a first edge pointing from the first node to the second node, and a second edge pointing from the second node to the first node. Each first node corresponds to a job position, each second node corresponds to a material, each first edge indicates the process action of the job position towards the material, and each second edge indicates the action input of the material towards the job position. The organizational structure sub-model includes multiple third nodes and third edges connecting each third node. Each third node corresponds to a job level of a job position, and each third edge indicates the job level action between two adjacent job levels. The business specification sub-model includes multiple fourth nodes, multiple fifth nodes, and fourth edges connecting each fourth node and its corresponding fifth node. Each fourth node corresponds to an execution subject of a business, each fifth node corresponds to an execution object of a business, and each fourth edge indicates the business specification of the connected execution subject towards the connected execution object.

[0008] Furthermore, in some embodiments of the present invention, the processing system further includes a business knowledge base. The business knowledge base includes knowledge concepts from the business process sub-model, the organizational structure sub-model, and the business specification sub-model, as well as the logical relationships between these knowledge concepts. The logical relationships include at least deductive relationships and attributive relationships.

[0009] Furthermore, in some embodiments of the present invention, the step of modifying the nodes and / or edges of at least one of the business process sub-model, the organizational structure sub-model, and the business specification sub-model according to the business data includes: parsing the business data and matching the obtained parsing results with the business process sub-model, the organizational structure sub-model, and the business specification sub-model and each of the nodes therein; and in response to the parsing result matching any one of the business process sub-model, the organizational structure sub-model, and the business specification sub-model, but not matching any of the nodes therein, creating a new node and an edge connecting the new node in the matched sub-model according to the parsing result.

[0010] Furthermore, in some embodiments of the present invention, the step of modifying the nodes and / or edges of at least one of the business process sub-model, the organizational structure sub-model, and the business specification sub-model according to the business data further includes: in response to the parsing result matching any of the nodes in the business process sub-model, the organizational structure sub-model, and the business specification sub-model, updating the node parameters of the matching node and / or updating the edge parameters of the edges connecting the node according to the parsing result.

[0011] Furthermore, in some embodiments of the present invention, the step of parsing the business data includes: parsing the business data according to preset semantic slots to determine at least one of time information, location information, person information, event information, trigger condition information, and method requirement information in the business data. The step of matching the obtained parsing results with the business process sub-model, the organizational structure sub-model, and the business specification sub-model, and each of the nodes therein, includes: semantically matching the time information, location information, person information, event information, trigger condition information, and method requirement information in the business data with the model description information of the business process sub-model, the organizational structure sub-model, and the business specification sub-model, as well as the node parameters of each node in each sub-model.

[0012] Furthermore, in some embodiments of the present invention, the first intelligent agent is further configured to: determine an initial value based on the importance of each node in the business process sub-model, the organizational structure sub-model, and the business specification sub-model; count the sending time and / or receiving time of business data involving each node, and determine an adjustment value based on the duration from the current time to the sending time and / or the receiving time; determine the activity value of each node based on the initial value and the adjustment value; and prune the node in response to the activity value of any node being lower than a preset activity threshold.

[0013] Furthermore, in some embodiments of the present invention, the processing system further includes at least one second agent configured at at least one node of the business process sub-model and / or the organizational structure sub-model, and configured to: generate corresponding second business data based on first business data provided by an upstream neighbor node adjacent to the local node where the second agent is located. The upstream neighbor node includes a first upstream neighbor node in the business process sub-model preceding the local node, a second upstream neighbor node in the organizational structure sub-model superior to the local node, and a third upstream neighbor node in the business specification sub-model adjacent to the local node; and provides the second business data to a downstream neighbor node adjacent to the local node. The downstream neighbor node includes a first downstream neighbor node in the business process sub-model following the local node, and / or a second downstream neighbor node in the organizational structure sub-model subordinate to the local node.

[0014] Furthermore, the above-mentioned business data processing method provided by the second aspect of the present invention includes the following steps: obtaining business data from multiple nodes of a business graph model, an organizational structure sub-model, and a business specification sub-model of a business data processing system as provided by the first aspect of the present invention; and modifying, based on the business data, at least one of the nodes and / or edges of the business process sub-model, the organizational structure sub-model, and the business specification sub-model.

[0015] Furthermore, the computer-readable storage medium provided according to the third aspect of the present invention stores computer instructions thereon. When the computer instructions are executed by a processor, a method for processing business data as provided in the second aspect of the present invention is implemented. Attached Figure Description

[0016] The above-described features and advantages of the present invention will be better understood after reading the following detailed description of embodiments of the present disclosure in conjunction with the accompanying drawings. In the drawings, components are not necessarily drawn to scale, and components having similar related characteristics or features may have the same or similar reference numerals.

[0017] Figure 1 A schematic diagram of a business processing system provided according to some embodiments of the present invention is shown.

[0018] Figure 2 A schematic diagram of the architecture of a business processing system provided according to some embodiments of the present invention is shown.

[0019] Figure 3 A schematic diagram of a business process sub-model provided according to some embodiments of the present invention is shown.

[0020] Figure 4 A schematic diagram of an organizational sub-model provided according to some embodiments of the present invention is shown.

[0021] Figure 5 A schematic diagram of a business specification sub-model provided according to some embodiments of the present invention is shown.

[0022] Figure 6 A flowchart illustrating a method for processing business data according to some embodiments of the present invention is shown. Detailed Implementation

[0023] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Although the description of the present invention is presented in conjunction with preferred embodiments, this does not mean that the features of the invention are limited to these embodiments. On the contrary, the purpose of describing the invention in conjunction with embodiments is to cover other options or modifications that may be derived based on the claims of the present invention. To provide a thorough understanding of the invention, many specific details will be included in the following description. The invention may also be implemented without using these details. Furthermore, to avoid confusion or obscuring the focus of the invention, some specific details will be omitted in the description.

[0024] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0025] Furthermore, the terms "upper," "lower," "left," "right," "top," "bottom," "horizontal," and "vertical" used in the following description should be understood as the orientations shown in the relevant paragraphs and accompanying drawings. These relative terms are for illustrative purposes only and do not imply that the described apparatus must be manufactured or operated in a specific orientation, and therefore should not be construed as limiting the invention.

[0026] It is understood that although terms such as "first," "second," and "third" may be used herein to describe various components, regions, layers, and / or parts, these components, regions, layers, and / or parts should not be limited by these terms, and these terms are only used to distinguish different components, regions, layers, and / or parts. Therefore, the first components, regions, layers, and / or parts discussed below may be referred to as second components, regions, layers, and / or parts without departing from some embodiments of the present invention.

[0027] As mentioned above, industrial-grade scenarios demand extremely high levels of reliability, security, and determinism. The probabilistic output characteristics of intelligent agents fundamentally conflict with the zero-fault-tolerance requirements of production lines. Furthermore, production systems involve complex processes such as equipment control, process parameters, and supply chain coordination, requiring strict adherence to physical laws and industry standards. Current Large Language Models (LLMs) still carry the risk of illusion; flawed decisions could lead to production line shutdowns, safety accidents, or substantial economic losses. Moreover, data silos and complex proprietary protocols are prevalent within enterprises, making it difficult for intelligent agents to acquire high-quality real-time data across systems. Additionally, production processes heavily rely on domain expert experience, and general-purpose models lack the accumulation of vertical industry knowledge. Furthermore, the three mainstream technical paradigms currently integrating large language models and knowledge graphs (KG)—LEK (LLM EnhancesKG, large language model enhances knowledge graph), KEL (KG Enhances LLM, knowledge graph enhances large language model), and LKC (LLM & KG Collaborative, bidirectional collaboration between large language model and knowledge graph)—while achieving automated knowledge graph construction, enhanced retrieval and suppression of model illusions, and bidirectional collaborative evolution of knowledge and model, are merely general-level technical integration paths. They cannot comprehensively represent the complete business operation mechanism within an enterprise, and are insufficient to construct an enterprise-level business knowledge ontology with strict permissions, behaviors, and governance constraints. Moreover, the current LKC, being based on a closed loop of self-generated knowledge, is prone to internal error loop propagation.

[0028] Specifically, at the knowledge construction level corresponding to the LEK paradigm, existing technologies mostly employ a single knowledge graph or rely solely on LLM for entity and relation extraction, making it difficult to distinguish different semantic levels such as business processes, organizational structures, specifications, and job knowledge. This invention constructs business data into business process sub-models, organizational structure sub-models, and business specification sub-models, and assigns business semantics to nodes and edges (e.g., job actions, permission settings, rule constraints, inference, and attribution relationships). This enables LLM to accurately identify and attach new concepts at different levels during data-driven expansion, avoiding the ontology drift and semantic confusion common in LEK. Simultaneously, through a pruning mechanism, redundant or erroneous nodes are removed using official documents and actual business activity data, solving the knowledge base pollution problem caused by LLM illusion in LEK and significantly improving the accuracy and stability of the knowledge structure.

[0029] Furthermore, at the knowledge application level corresponding to the KEL paradigm, existing KELs typically only use knowledge graphs as static fact sources for retrieval enhancement, lacking fine-grained control over the association between permissions, behavioral norms, and dynamic tasks. This invention, through a multi-layer graph model-based association retrieval mechanism, uses job positions / job titles as core anchors to organically link business processes, organizational structures, and business norms. This allows intelligent agents to not only acquire relevant knowledge corresponding to KEL functions (KEL functions) when performing tasks, but also simultaneously obtain permission boundaries, behavioral rules, and method paths. This achieves integrated reasoning that integrates process-driven execution, organizationally defined permissions, normatively constrained behavior, and knowledge-supported decision-making, overcoming the risks of unauthorized access or violations arising from the lack of contextual constraints in traditional KELs in complex business scenarios.

[0030] Furthermore, at the knowledge evolution level corresponding to the LKC paradigm, existing LKCs mostly remain at the academic level of two-way collaboration, lacking closed-loop integration with actual business operations. This invention constructs a data-driven two-way evolution mechanism through data expansion and pruning mechanisms. On one hand, LLM identifies new people, events, and implicit logic from specific task data, dynamically expanding nodes and edges. On the other hand, through business activity statistics and management document feedback, the graph model is pruned and its structure adjusted, forming a continuous closed loop of "data → graph model → agent execution → new data". This mechanism not only retains the collaborative advantages of LKC but also introduces business effectiveness verification, avoiding the knowledge inflation and accumulation of failures caused by the lack of governance in purely academic LKCs.

[0031] Furthermore, at the multi-agent collaboration level, existing technologies typically rely on pre-defined interaction protocols or centralized scheduling, making it difficult to adaptively adjust collaborative relationships based on business changes. This invention presents a multi-agent autonomous collaboration mechanism based on a business knowledge model. It directly maps the business positioning of each agent (e.g., job title, position, task outcome) to a multi-layer graph model, enabling it to autonomously identify process and organizational relationships through the graph structure and dynamically adjust collaboration paths without additional manual configuration. This graph model-based autonomous collaboration mechanism significantly reduces the maintenance cost of multi-agent systems and improves response speed and adaptability during business changes.

[0032] To overcome the aforementioned deficiencies in the existing technology, this invention provides a business data processing system, a business data processing method, and a computer-readable storage medium. By constructing a multi-level composite graph model and modifying the nodes and / or edges of multiple sub-models in the composite graph model according to the acquired business data, it can effectively improve the reliability, security, and determinism of intelligent agents performing tasks in the production field.

[0033] Specifically, this invention, through the hierarchical definition of a multi-level composite graph model, association retrieval, data-driven expansion and pruning, and an agent autonomous collaboration mechanism, produces practical effects that are superior to the existing LEK / KEL / LKC paradigms in four aspects: accuracy of knowledge construction, compliance of application, effectiveness of evolution, and adaptability of multi-agent collaboration.

[0034] In some non-limiting embodiments, the business data processing system provided in the first aspect of the present invention can be implemented based on the business data processing method provided in the second aspect of the present invention.

[0035] Please refer to the reference for details. Figure 1 and Figure 2 . Figure 1 A schematic diagram of a business processing system provided according to some embodiments of the present invention is shown. Figure 2 A schematic diagram of the architecture of a business processing system provided according to some embodiments of the present invention is shown.

[0036] exist Figure 1 and Figure 2 In the illustrated embodiment, the business data processing system provided by the first aspect of the present invention includes a business graph model and a first intelligent agent. Here, the business graph model includes a business process sub-model, an organizational structure sub-model, and a business specification sub-model, wherein each of the business process sub-model, organizational structure sub-model, and business specification sub-model includes multiple nodes and edges connecting each node.

[0037] Specifically, the knowledge learning process is as follows: Figure 1 As shown, all incremental information data in the enterprise (including information status on business systems, work dialogues between employees, internal documents of the enterprise, etc.) will be input into LLM (Large Language Model). LLM will perform structured parsing of the input increments according to the structure of the graph model, and the parsing results will be given to the graph model for model expansion. The newly expanded information will be used for concept clarification in the knowledge base.

[0038] In addition, the knowledge application process, such as Figure 2 As shown, a task first enters the graph model for task localization, thereby enriching task information (including the corresponding job position, upstream and downstream linkages, output requirements, execution basis, input dependencies, etc.). Then, the first agent, used for task planning, decomposes the task and sends the relevant sub-task information to the second agent corresponding to the job position. After receiving the sub-task, the second agent first refers to the graph model and knowledge base to understand the task, completes the sub-task as required, and then interacts and collaborates with the second agent as instructed.

[0039] Please refer to the reference. Figures 3-5 . Figure 3 A schematic diagram of a business process sub-model provided according to some embodiments of the present invention is shown. Figure 4 A schematic diagram of an organizational sub-model provided according to some embodiments of the present invention is shown. Figure 5 A schematic diagram of a business specification sub-model provided according to some embodiments of the present invention is shown.

[0040] exist Figure 3 In the embodiment shown, the business process sub-model includes multiple first nodes, multiple second nodes, a first side pointing from the first node to the second node, and a second side pointing from the second node to the first node. Each first node corresponds to a job position, each second node corresponds to a material, each first side indicates the job position's process action on the material, and each second side indicates the material's action input to the job position.

[0041] Here, the aforementioned process needs to be linked to the company's normal business flow, action specifications, and triggering conditions. The process actions represent the daily work tasks of a position, and each position will have a corresponding outcome for each task event. This outcome corresponds to one or more materials, which may be the input for its own subsequent work (i.e., an edge connecting itself to the current position's node), or it may serve as the input for other positions' work (i.e., an edge connecting the target node to the current position's node).

[0042] For example, a testing company typically includes positions such as sales, technology, customer service, warehousing, laboratory, report writing, and finance. The workflow for each position includes: sales first connects with the client to discuss testing needs; technology analyzes the needs and provides testing solutions and quotations; customer service receives samples from clients and sorts and verifies them; warehousing stores the sorted samples according to demand; laboratory personnel retrieve samples for testing according to the project schedule; report writers summarize the experimental results to create a client report; and finance liaises with clients to confirm project payments, etc.

[0043] exist Figure 4 In the embodiment shown, the organizational structure sub-model includes multiple third nodes and third sides connecting each third node. Each third node corresponds to a job level, and each third side indicates the job level action between two adjacent job levels.

[0044] Here, each third node is associated with the first node in the business process sub-model, and correspondingly with an employee ID. Job-level actions include, for example, a superior assigning a task to a subordinate, or a subordinate reporting their work to a superior.

[0045] For example, each position includes two levels: general staff and business manager. The business manager will assign business projects to the general staff, and the general staff will report to the business manager after completing their tasks.

[0046] exist Figure 5 In the embodiment shown, the business specification sub-model includes multiple fourth nodes, multiple fifth nodes, and fourth edges connecting each fourth node and its corresponding fifth node. Each fourth node corresponds to an execution subject of a business, each fifth node corresponds to an execution object of a business, and each fourth edge indicates the business specification of the connected execution subject to the connected execution object.

[0047] Here, the business specification sub-model mainly expresses, based on the parsed specification concept, "who, under what conditions, to what object, according to what rules, performs what action, and bears what consequences." The executing entity corresponds to... Figure 3 or Figure 4 The position or rank in the job.

[0048] For example, assuming there are review standards, the sales manager needs to collaborate with the technical and customer service managers to review the testing plan.

[0049] Thus, the business data processing system, method, and storage medium provided by the present invention can adopt an overall strategy of taking job position or rank as the core anchor point, business object as the flow link, and rules and knowledge as the two-way empowerment, to realize the organic connection of the sub-models of the three-layer business graph model, and form a three-layer linkage system of process-driven execution, organization-limited authority, and standardized behavior constraint.

[0050] First, the business process sub-model serves as the main operating framework, while the organizational structure sub-model provides hierarchical and permission constraints. It links the "job nodes" in the process to the "job level nodes" in the organizational structure diagram, and adds the approval and reporting paths of the corresponding positions to the process side, thereby achieving the unification of "tasks, people, and authority".

[0051] Secondly, a business specification sub-model is introduced as a behavioral constraint layer. The executing subjects in the sub-model are aligned with process positions and organizational ranks, and the executing objects are aligned with the tasks, events, and outputs in the process. Trigger conditions and rule references are marked by edge attributes, so that each process edge can be traced back to specific specification clauses and consequence definitions.

[0052] Please continue to refer to this. Figure 1 .like Figure 1 As shown, all incremental information and data within the enterprise is input into the LLM (Logistic Model). The LLM performs structured parsing of the input increments based on the graph model's structure. The parsing results are then fed back into the graph model for model expansion. The newly expanded information undergoes conceptual clarification in the knowledge base. Here, the enterprise's information and data include the status of information in business systems, work conversations between employees, and internal enterprise documents.

[0053] In addition, Figure 1In the illustrated embodiment, the business data processing system provided by the first aspect of the present invention further includes a business knowledge base. Here, the business knowledge base includes knowledge concepts from the business process sub-model, organizational structure sub-model, and business specification sub-model.

[0054] Here, the business knowledge base primarily focuses on clarifying and storing raw materials related to all concepts in the graphical model, such as various positions, job levels, actions, and standards. Clarification includes, for example, defining a position (work objectives, job content, etc.), job levels (authority, responsibility, management scope, etc.), actions (action process description, subject / object description), and various standards (scope of application, objects of application, etc.).

[0055] For example, a business knowledge base in the testing field typically includes knowledge concepts of various national standards, enterprise standards, and other standard documents, as well as knowledge concepts of historical projects.

[0056] Thus, a business knowledge base is embedded as a supporting layer. By establishing associations between knowledge concept nodes and task events, the required business knowledge is dynamically associated during process execution through the deduction and attribution relationships of the knowledge graph, thereby achieving a closed loop from "doing tasks" to "understanding rules and methods".

[0057] In addition, Figure 1 In the embodiments shown, the business data processing system provided by the first aspect of the present invention further includes a second intelligent agent, which is configured at at least one node of the business process sub-model and / or the organizational structure sub-model, for receiving the first business data output by the upstream neighbor nodes in the three types of sub-models of business process, organizational structure and business knowledge, generating the second business data, and then sending the generated second business data to the downstream nodes of the business process sub-model and / or the lower-level nodes of the organizational structure sub-model.

[0058] Here, the knowledge application process is as follows: Figure 1 As shown, a task first enters the graph model for task localization, thereby enriching the task information. This task information includes the corresponding job position, the upstream and downstream linkages of the task, the task's output requirements, execution basis, and input dependencies. Next, the first intelligent agent decomposes the task and sends the relevant sub-task information to the corresponding second intelligent agents. After receiving the sub-tasks, the second intelligent agent first refers to the graph model and knowledge base to understand the task, completes the sub-tasks as required, and then interacts and collaborates with other second intelligent agents as instructed.

[0059] In some non-limiting embodiments, the business data processing system provided in the first aspect of the present invention includes a memory and a controller. Here, the memory includes, but is not limited to, the computer-readable storage medium provided in the third aspect above, on which computer instructions are stored. The controller is connected to the memory and configured to execute the computer instructions stored in the memory to implement the business data processing method as provided in the second aspect of the present invention.

[0060] The working principle of the above-mentioned business data processing system will be described below with reference to some embodiments of business data processing methods. Those skilled in the art will understand that these embodiments of business data processing methods are merely non-limiting implementations provided by the present invention, intended to clearly demonstrate the main concepts of the invention and provide specific solutions convenient for public implementation, rather than limiting all functions or all working methods of the business data processing system. Similarly, the business data processing system is also only one non-limiting implementation provided by the present invention, and does not limit the executing entity or execution order of the steps in these business data processing methods.

[0061] Please refer to Figure 6 . Figure 6 A flowchart illustrating a method for processing business data according to some embodiments of the present invention is shown.

[0062] like Figure 6 As shown, the first intelligent agent can first obtain business data from multiple nodes of the business process sub-model, organizational structure sub-model, and business specification sub-model of the above business graph model.

[0063] Here, the aforementioned business data refers to specific task data in business operations, including the initiator and task content. The initiator will inevitably correspond to the job position node and grade node in the graph model. Task content typically includes time (when), location (where), person (who), event (what), triggering condition (why), and method requirements (how).

[0064] After that, as Figure 6 As shown, the first intelligent agent can modify at least one of the nodes and / or edges in the business process sub-model, organizational structure sub-model, and business specification sub-model based on the aforementioned business data.

[0065] Specifically, during the above modification process, the first intelligent agent can first parse the business data and then match the obtained parsing results with the business process sub-model, the organizational structure sub-model, and the business specification sub-model, as well as each node therein.

[0066] Furthermore, during the process of parsing business data, the first intelligent agent can parse the business data according to the preset semantic slots to determine at least one of the following information in the business data: time information, location information, person information, event information, trigger condition information, and method requirement information.

[0067] During the matching process, the first intelligent agent can semantically match the time information, location information, personnel information, event information, triggering condition information, and method requirement information in the business data with the model description information of the business process sub-model, organizational structure sub-model, and business specification sub-model, as well as the node parameters of each node in each sub-model.

[0068] Here, the first intelligent agent can calculate the semantic similarity between the aforementioned time information, location information, personnel information, event information, triggering condition information, and method requirement information and the business process sub-model, organizational structure sub-model, and business specification sub-model, as well as all nodes therein, and determine the job node with the highest semantic similarity as the target node.

[0069] Those skilled in the art will understand that the above embodiments for determining target nodes by calculating semantic similarity are merely some non-limiting implementations provided by the present invention, intended to clearly demonstrate the main concept of the present invention and provide some specific solutions that are easy for the public to implement, rather than intended to limit the scope of protection of the present invention.

[0070] Optionally, in other embodiments, the first intelligent agent may also extract the text features corresponding to the time information, location information, person information, event information, trigger condition information and method requirement information mentioned above, perform unified vector encoding, and complete node matching through the cosine similarity between the encoded vectors.

[0071] Optionally, in other embodiments, the first intelligent agent can also be replaced by matching methods such as keyword weight matching, text edit distance calculation, graph node structural feature weighted scoring, and large model direct inference filtering, and select the node with the best matching degree as the target node.

[0072] Subsequently, in response to the parsing result matching any one of the business process sub-model, organizational structure sub-model, and business specification sub-model, but not matching any node, the first agent can create a new node and an edge connecting the new node in the matching sub-model based on the parsing result.

[0073] For example, when the LLM identifies a newly emerging position, such as a temporarily added position, the first agent can match it to the business process sub-model, but not to any existing node within it. In this case, the first agent can create a new node at the corresponding level and attach it to the aforementioned target node, automatically completing the business process chain.

[0074] For example, when the LLM identifies a newly emerging job level, such as a temporarily added job level, the first agent can match it to the organizational structure sub-model, but not to any existing node within it. In this case, the first agent can create a new node at the corresponding level and attach it to the aforementioned target node, automatically completing the organizational structure affiliation.

[0075] For example, when the LLM identifies a newly emerging business scenario, such as temporarily involving a new business element, the first agent can match it to the business specification sub-model, but not to any existing node within it. In this case, the first agent can create a new node at the corresponding level and attach it to the aforementioned target node, automatically completing the business specification constraints.

[0076] Furthermore, during the modification process, in response to the parsing result matching any node in the business process sub-model, organizational structure sub-model, and business specification sub-model, the first agent can update the node parameters of the matching node and / or update the edge parameters of the edges connecting the nodes based on the parsing result.

[0077] Specifically, updating node parameters includes rewriting the parameters of the matching node, clearing all node parameters corresponding to the matching node, and deleting the matching node. Updating edge parameters includes rewriting the parameters of existing edges connecting to the matching node, removing one or more existing edges connecting to the matching node to sever the association between the corresponding nodes, and creating new associated edges with the matching node as the endpoint to establish an association between the node and other target nodes. Here, various node parameter update operations and various edge parameter update operations can be executed independently or in combination synchronously based on the parsing results.

[0078] For the "method requirements (how)" and "triggering conditions (why)," LLM extracts the implicit logic and transforms it into new "deductive" relationships or new "constraint" edges in the knowledge graph or canonical graph. For example, if a specific compliance judgment logic repeatedly appears in the task data, the first agent will automatically generate a new deductive path from "project plan" to "compliance judgment" and update the weights of the corresponding job knowledge nodes. Through this continuous data feedback and semantic parsing, the first agent enables the business graph model to leap from a static architecture to a dynamic cognitive network with self-evolution capabilities.

[0079] In addition, in some implementations, the first intelligent agent can also determine an initial value based on the importance of each node in the business process sub-model, organizational structure sub-model, and business specification sub-model.

[0080] Then, the first intelligent agent can count the sending and / or receiving times of the business data involved in each node, and determine an adjustment value based on the duration from the current time to the sending and / or receiving time.

[0081] Then, the first agent can determine the activity value of each node based on the initial value and the adjustment value.

[0082] Subsequently, in response to any node's activity value falling below a preset activity threshold, the first agent can perform pruning on the node.

[0083] Those skilled in the art will understand that the above-described embodiments of statistical node activity based on time weighting are merely some non-limiting implementations provided by the present invention, intended to clearly demonstrate the main concept of the present invention and provide some specific solutions that are easy for the public to implement, rather than intended to limit the scope of protection of the present invention.

[0084] Optionally, in other embodiments, the first intelligent agent may also determine the activity level of each node based on the frequency of event triggering. Specifically, the first intelligent agent may count the number of business process instances participated in by each node, the number of times it was accessed or called, and the hit frequency as a critical path node in decision-making within a preset historical time period. Then, the first intelligent agent may determine a comprehensive trigger score based on the cumulative number of triggers for each node within the historical time period, combined with the business weight of each trigger. Subsequently, the first intelligent agent may determine the activity value of each node based on the comprehensive trigger score and the node's hierarchical coefficient in the organizational structure. In response to any node's activity value falling below a preset activity threshold, the first intelligent agent may prune the node.

[0085] Specifically, the pruning process can gradually remove long-term inactive business data edges according to a preset time interval, based on the edge activity value from low to high. The node itself is removed only after all its associated edges have been pruned. Furthermore, the first intelligent agent can differentiate the activity threshold and pruning granularity based on the importance level of the business domain to which the node belongs, retaining a minimum number of associated edges for core business nodes to avoid accidentally deleting critical business topologies.

[0086] Thus, the business data processing system, method, and storage medium provided by the present invention can be updated with business knowledge through the above modifications and pruning processes.

[0087] For example, when a sales representative receives a new requirement from a customer regarding "low-temperature cycle performance testing of lithium batteries," the first agent analyzes the task content and identifies "low-temperature cycle testing," a new business event not defined in the existing business process diagram. It then adds a "low-temperature cycle testing" node to the business process diagram and attaches it to the "battery testing" parent category through semantic similarity calculation. Simultaneously, it adds the corresponding national standard GB / T XXX as a constraint edge to the specification diagram and adds knowledge points such as "low-temperature testing equipment selection" and "temperature control accuracy requirements" and their derivation relationships to the job knowledge graph. Subsequently, the new parameter settings and result judgment logic generated by the experimental team during the task execution are extracted by LLM and enriched as new attributes to the "experimental method" node. Meanwhile, the first agent periodically tracks the activity of each node based on a pruning mechanism. If an older version of a testing standard has no task calls for three consecutive months, it is marked as a dormant node and its search weight by the agent is reduced. When the company released a new version of the "Review Management Measures" which explicitly canceled the collaborative review process for sales managers, the first intelligent agent immediately removed the corresponding "collaborative review" edge from the specification diagram based on the data in the official document, thereby achieving dynamic and precise evolution of business knowledge.

[0088] Furthermore, in some embodiments, the business data processing system provided by the present invention can also complete the autonomous collaborative work of multiple agents via at least one second agent.

[0089] Specifically, please refer to the following: Figures 1-5 .like Figures 1-5 As shown, the second agent can generate corresponding second business data based on the first business data provided by the upstream neighbor nodes adjacent to the local node where the second agent is located. Here, the upstream neighbor nodes include the first upstream neighbor node located before the local node in the business process sub-model, the second upstream neighbor node located above the local node in the organizational structure sub-model, and the third upstream neighbor node adjacent to the local node in the business specification sub-model.

[0090] Subsequently, the second agent can provide second business data to downstream neighbor nodes adjacent to the local node. Here, downstream neighbor nodes include the first downstream neighbor node located after the local node in the business process sub-model, and / or the second downstream neighbor node located below the local node in the organizational structure sub-model.

[0091] For example, each job position corresponds to a second intelligent agent. Upon initialization, each second intelligent agent can retrieve its predecessor or successor node in the business process diagram from the business process sub-model, knowing who its upstream is and who its downstream task is. Furthermore, it can retrieve its superior or subordinate within the organizational structure from the organizational structure sub-model, knowing who it reports to and who assigns it tasks. It can also retrieve the review rules and consequences to be followed from the business specification sub-model, such as the need for a testing plan to be reviewed by the technical manager, customer service manager, and sales manager. Finally, it can retrieve related standard clauses and deductive logic from the job knowledge sub-model, such as project plan + specification → compliance judgment.

[0092] For a single testing commission process, the multi-agent autonomous collaboration generally proceeds as follows: After the customer submits the testing commission, the sales agent automatically identifies subsequent steps based on the business process diagram and sends the structured requirements to the technical agent, without the need for pre-setting routing rules. The technical agent then queries the business specification diagram and finds that the testing solution has triggered collaborative review conditions due to exceeding the amount limit. It then autonomously contacts the sales manager, technical manager, and customer service manager for parallel review, and only after confirmation from all parties is the final solution locked in. After the solution is determined, the technical agent notifies the customer service agent to receive the samples, and the customer service agent then locates the warehouse agent through the process diagram to complete the warehousing registration. After the warehouse agent confirms that the samples are ready, it notifies the experimental agent. Before execution, the experimental agent retrieves the corresponding national standards, equipment parameters, and compliance derivation logic from the job knowledge graph to guide the testing and writes the experimental results back to the graph model. After the reporting agent detects that the task status has changed to "experiment completed," it automatically captures data, generates a testing report, and delivers it to the customer. At the same time, it triggers the financial agent to start the payment collection process. The entire commission process is driven by a multi-level composite graph model, which enables autonomous connection and compliant collaboration among the various agents.

[0093] In summary, the business data processing system, business data processing method, and computer-readable storage medium provided by this invention can all effectively improve the reliability, security, and determinism of intelligent agents performing tasks in the production field by constructing multi-level composite graph models and modifying the nodes and / or edges of multiple sub-models in the composite graph model according to the acquired business data.

[0094] Although the methods described above are illustrated and depicted as a series of actions for the sake of simplicity, it should be understood and appreciated that these methods are not limited by the order of the actions, as some actions may occur in a different order and / or concurrently with other actions from the illustrations and descriptions herein or not illustrated and described herein but which may be understood by those skilled in the art, according to one or more embodiments.

[0095] Those skilled in the art will understand that information, signals, and data can be represented using any of a variety of different techniques and skills. For example, the data, instructions, commands, information, signals, bits, symbols, and chips described throughout the above description can be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, light fields or optical particles, or any combination thereof.

[0096] Those skilled in the art will further appreciate that the various illustrative logic blocks, modules, circuits, and algorithm steps described in conjunction with the embodiments disclosed herein can be implemented as electronic hardware, computer software, or a combination of both. To clearly illustrate this interchangeability between hardware and software, the various illustrative components, blocks, modules, circuits, and steps are described above in a generalized manner in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system. Those skilled in the art may implement the described functionality in different ways for each specific application, but such implementation decisions should not be construed as departing from the scope of the invention.

[0097] The various illustrative logic modules and circuits described in conjunction with the embodiments disclosed herein may be implemented or performed using a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. The general-purpose processor may be a microprocessor, but in alternatives, it may be any conventional processor, controller, microcontroller, or state machine. The processor may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors cooperating with a DSP core, or any other such configuration.

[0098] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of both. The software module may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to a processor such that the processor can read and write information to / from the storage medium. In an alternative, the storage medium may be integrated into the processor. The processor and storage medium may reside in an ASIC. The ASIC may reside in a user terminal. In an alternative, the processor and storage medium may reside as discrete components in the user terminal.

[0099] In one or more exemplary embodiments, the described functionality may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software as a computer program product, the functionality may be stored or transmitted as one or more instructions or code on or through a computer-readable medium. A computer-readable medium includes both computer storage media and communication media, encompassing any medium that facilitates the transfer of a computer program from one location to another. A storage medium may be any available medium accessible to a computer. By way of example and not limitation, such a computer-readable medium may include RAM, ROM, EEPROM, CD-ROM or other optical disc storage, disk storage or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and is accessible to a computer. Any connection is also legitimately referred to as a computer-readable medium. For example, if the software is transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of a medium. As used in this article, disk and disc include compact discs (CDs), laser discs, optical discs, digital multi-purpose discs (DVDs), floppy disks, and Blu-ray discs. Disks typically reproduce data magnetically, while discs reproduce data optically using lasers. Combinations of these should also be included within the scope of computer-readable media.

[0100] The prior description of this disclosure is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to this disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not intended to be limited to the examples and designs described herein, but should be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A business data processing system, characterized in that, include: The business graph model includes a business process sub-model, an organizational structure sub-model, and a business specification sub-model, wherein each of the business process sub-model, the organizational structure sub-model, and the business specification sub-model includes multiple nodes and edges connecting each node. as well as A first intelligent agent is configured to: acquire business data from each of the nodes of the business process sub-model, the organizational structure sub-model, and the business specification sub-model; and modify the nodes and / or edges of at least one of the business process sub-model, the organizational structure sub-model, and the business specification sub-model based on the business data.

2. The processing system as described in claim 1, characterized in that, The business process sub-model includes multiple first nodes, multiple second nodes, a first edge pointing from the first node to the second node, and a second edge pointing from the second node to the first node. Each first node corresponds to a job position, and each second node corresponds to a material. Each first edge indicates the job position's process action on the material, and each second edge indicates the material's action input to the job position, and / or The organizational structure sub-model includes multiple third nodes and third edges connecting each third node. Each third node corresponds to a job level for one of the job positions, and each third edge indicates the job level action between two adjacent job levels, and / or The business specification sub-model includes multiple fourth nodes, multiple fifth nodes, and fourth edges connecting each fourth node and its corresponding fifth node. Each fourth node corresponds to an execution subject of the business, each fifth node corresponds to an execution object of the business, and each fourth edge indicates the business specification of the connected execution subject to the connected execution object.

3. The processing system as described in claim 2, characterized in that, It also includes a business knowledge base, which contains knowledge concepts from the business process sub-model, the organizational structure sub-model, and the business specification sub-model.

4. The processing system as described in claim 2, characterized in that, The step of modifying at least one of the nodes and / or edges in the business process sub-model, the organizational structure sub-model, and the business specification sub-model based on the business data includes: The business data is parsed, and the parsing results are matched with the business process sub-model, the organizational structure sub-model, and the business specification sub-model, as well as each of the nodes therein; and In response to the parsing result matching any one of the business process sub-model, the organizational structure sub-model, and the business specification sub-model, but not matching any of the nodes, a new node and an edge connecting the new node are created in the matching sub-model according to the parsing result.

5. The processing system as described in claim 4, characterized in that, The step of modifying at least one of the nodes and / or edges in the business process sub-model, the organizational structure sub-model, and the business specification sub-model based on the business data further includes: In response to the parsing result matching any of the nodes in the business process sub-model, the organizational structure sub-model, and the business specification sub-model, the node parameters of the matching node are updated, and / or the edge parameters of the edges connecting the node are updated, based on the parsing result.

6. The processing system as described in claim 4, characterized in that, The step of parsing the business data includes: parsing the business data according to preset semantic slots to determine at least one of the following in the business data: time information, location information, person information, event information, trigger condition information, and method requirement information. The step of matching the obtained parsing results with the business process sub-model, the organizational structure sub-model, and the business specification sub-model and each of the nodes therein includes: semantically matching the time information, location information, person information, event information, triggering condition information, and method requirement information in the business data with the model description information of the business process sub-model, the organizational structure sub-model, and the business specification sub-model, as well as the node parameters of each node in each sub-model.

7. The processing system as described in claim 1, characterized in that, The first intelligent agent is also configured as follows: An initial value is determined based on the importance of each node in the business process sub-model, the organizational structure sub-model, and the business specification sub-model. The sending and / or receiving times of the business data involved in each of the nodes are statistically analyzed, and an adjustment value is determined based on the duration from the current time to the sending and / or receiving time. Based on the initial value and the adjustment value, the activity value of each node is determined; as well as If the activity value of any of the nodes is lower than a preset activity threshold, the node is pruned.

8. The processing system as described in claim 1, characterized in that, It also includes at least one second intelligent agent, which is configured at at least one of the nodes of the business process sub-model and / or the organizational structure sub-model, and is configured to: Based on the first business data provided by the upstream neighbor nodes adjacent to the local node where the second intelligent agent is located, corresponding second business data is generated. The upstream neighbor nodes include the first upstream neighbor node located before the local node in the business process sub-model, the second upstream neighbor node located above the local node in the organizational structure sub-model, and the third upstream neighbor node adjacent to the local node in the business specification sub-model; and The second business data is provided to downstream neighbor nodes adjacent to the local node, wherein the downstream neighbor nodes include the first downstream neighbor node located after the local node in the business process sub-model, and / or the second downstream neighbor node located below the local node in the organizational structure sub-model.

9. A method for processing business data, characterized in that, Includes the following steps: Business data is obtained from multiple nodes of the business process sub-model, organizational structure sub-model, and business specification sub-model of the business graph model of the business data processing system as described in any one of claims 1 to 8; as well as Based on the business data, modify at least one of the nodes and / or edges in the business process sub-model, the organizational structure sub-model, and the business specification sub-model.

10. A computer-readable storage medium storing computer instructions thereon, characterized in that, When the computer instructions are executed by the processor, the business data processing method as described in claim 9 is implemented.

Images