An automatic link building and dropping method and system based on agent cooperation

Abstract

The present application relates to the technical field of artificial intelligence and industry chain analysis, and discloses an automatic chain building and falling method and system based on agent cooperation, comprising: receiving the automatic chain building demand of a user, performing a first chain building operation on an industry chain directed acyclic graph by using an industry chain collaboration system based on the automatic chain building demand, and obtaining a first node; performing a subdivision chain building operation on the first node by using the industry chain collaboration system, and obtaining a second node, a third node and a leaf node; performing an industry chain labeling operation on the first node, the second node, the third node and the leaf node by using the industry chain collaboration system, and obtaining an industry chain labeling result; and performing an enterprise chain falling operation on the leaf node by using the industry chain collaboration system based on the industry chain labeling result, and obtaining a chain falling relationship database and an industry chain panoramic graph. The present application can solve the problems of dynamic chain building difficulty, multi-stage collaboration deficiency and low enterprise chain falling precision.

Description

Technical Field

[0001] This invention relates to an automatic chain building and chain termination method and system based on intelligent agent collaboration, belonging to the field of artificial intelligence and industrial chain analysis technology. Background Technology

[0002] Current supply chain analysis mainly relies on expert experience and manual surveys, which suffers from low efficiency, limited coverage dimensions, and lagging dynamic updates. Traditional methods use static database matching, which cannot combine dynamic factors such as policies and patents to generate supply chain maps. Generalization ability is based on multi-task multi-agent models and human-computer interaction. Single-agent solutions can only execute fixed tasks and lack task decomposition and multi-stage collaboration capabilities. Enterprise chain identification relies on keyword fuzzy matching, which does not integrate the three-dimensional features of technology, products, and supply chain, resulting in insufficient accuracy.

[0003] Therefore, existing technologies, due to the lack of static matching, fixed execution of single intelligent agents, and fusion of three-dimensional features, result in the inability to dynamically update the industry chain map, loss of control over multi-stage task collaboration, and insufficient accuracy in enterprise chain implementation. Summary of the Invention

[0004] This invention provides an automatic chain establishment and termination method and system based on intelligent agent collaboration. Its main purpose is to solve the problems of difficulty in dynamic chain establishment, lack of multi-stage collaboration, and low chain termination accuracy for enterprises.

[0005] To achieve the above objectives, this invention provides an automatic chain establishment and termination method based on agent collaboration, comprising:

[0006] Deploy a supply chain collaboration system, which includes a set of intelligent agents, a set of task stages, a directed acyclic graph of the supply chain, and a set of key-value pairs;

[0007] Receive the user's automatic chain building request, and based on the automatic chain building request, use the industrial chain collaboration system to perform a first-level chain building operation on the directed acyclic graph of the industrial chain to obtain a first-level node;

[0008] The supply chain collaboration system is used to perform subdivision chain building operations on the first-level nodes to obtain second-level nodes, third-level nodes, and leaf nodes.

[0009] The supply chain collaboration system is used to perform supply chain labeling operations on the first-level nodes, the second-level nodes, the third-level nodes, and the leaf nodes to obtain supply chain labeling results;

[0010] Based on the industry chain labeling results, the industry chain collaboration system is used to perform enterprise chain placement operations on the leaf nodes to obtain a chain placement relationship database and an industry chain panoramic map.

[0011] Optionally, the deployment of the supply chain collaboration system includes:

[0012] Initialize a set of intelligent agents, wherein the set of intelligent agents includes a planning intelligent agent, a scheduling intelligent agent, and an execution intelligent agent;

[0013] An initial task phase set is established, which includes primary chain building tasks, subdivided chain building tasks, industry chain labeling tasks, and enterprise chain placement tasks.

[0014] Initialize a directed acyclic graph of supply chains, wherein the directed acyclic graph of supply chains includes a set of empty nodes and a set of empty edges;

[0015] Establish a set of key-value pairs, wherein the set of key-value pairs includes a node context primary key prefix and a task intermediate result primary key prefix;

[0016] The supply chain collaboration system is determined by the set of intelligent agents, the set of task stages, the directed acyclic graph of the supply chain, and the set of key-value pairs.

[0017] Optionally, after deploying the supply chain collaboration system, it also includes:

[0018] The system is controlled to enter a state where it awaits user input.

[0019] Optionally, based on the automatic chain-building requirement, the step of using the industry chain collaboration system to perform a first-level chain-building operation on the directed acyclic graph of the industry chain to obtain first-level nodes includes:

[0020] Obtain the set of intelligent agents and the set of task stages in the industrial chain collaboration system; obtain the planning intelligent agent, scheduling intelligent agent and execution intelligent agent in the set of intelligent agents; obtain the first-level chain building task, subdivided chain building task, industrial chain labeling task and enterprise chain landing task in the set of task stages.

[0021] The planning agent is used to analyze the automatic chain building requirements to generate a task dependency graph consisting of the first-level chain building task, the subdivided chain building task, the industry chain labeling task, and the enterprise chain landing task;

[0022] According to the topological order in the task dependency graph, the scheduling agent is used to activate the first-level chain building task so as to control the first-level chain building task to be in a pending execution state.

[0023] While activating the primary chain building task, the primary chain building instruction of the primary chain building task is generated using the automatic chain building requirement, wherein the primary chain building instruction includes the region name and the industry name;

[0024] After receiving the first-level chain-building instruction, the execution agent invokes the graph-building tool.

[0025] The region name and industry name in the first-level chain building instruction are used as input parameters for the mapping tool;

[0026] By jointly querying knowledge bases and publicly available information on the Internet, the first-level nodes corresponding to the input parameters are output through the mapping tool.

[0027] Optionally, after performing a first-level chain building operation on the directed acyclic graph of the industrial chain based on the automatic chain building requirement and obtaining the first-level nodes using the industrial chain collaboration system, the process further includes:

[0028] Obtain the upstream and downstream edges, metadata, industry attributes, and constraint information of the first-level nodes connected by the mapping tool;

[0029] The industry attributes and the constraint information are written into the attribute fields and constraint fields of the structured data container of each first-level node in the first-level node, respectively, to obtain the updated structured data container.

[0030] The first-level node and the upstream and downstream edges are incorporated into the directed acyclic graph of the industrial chain to obtain the updated directed acyclic graph of the industrial chain.

[0031] Based on the primary key prefix of the intermediate results of the task, the intermediate results containing the first-level node, the upstream and downstream edges, the updated structured data container and the metadata are stored in a key-value pair set;

[0032] The updated directed acyclic graph of the industrial chain, which includes the first-level node and the upstream and downstream edges, and the updated structured data container are visualized to obtain a visualization interface.

[0033] Detect whether the user has triggered a confirmation operation on the visual interface;

[0034] After the confirmation operation is triggered, the pending state of the first-level chain building task is changed to the completed state, and the first-level node and the upstream and downstream edges are locked.

[0035] Optionally, the step of using the supply chain collaboration system to perform subdivision chain building operations on the first-level nodes to obtain second-level nodes, third-level nodes, and leaf nodes includes:

[0036] After the scheduling agent detects that the primary chain building task is in a completed state, it activates the sub-chain building task to control the sub-chain building task to be in a pending execution state.

[0037] The execution agent traverses the locked first-level nodes in the updated directed acyclic graph of the industry chain to call the updated structured data container corresponding to the first-level node.

[0038] Inherit the constraint information from the updated structured data container;

[0039] Using the first-level node as the parent node, the execution agent calls the mapping tool to query business information, patent clustering results, and supply chain data;

[0040] The business registration information, the patent clustering results, and the supply chain data are generated as secondary nodes, tertiary nodes, and leaf nodes.

[0041] Optionally, after performing subdivision and chain-building operations on the first-level nodes using the supply chain collaboration system to obtain second-level nodes, third-level nodes, and leaf nodes, the method further includes:

[0042] Business information, patent clustering results, and supply chain data are written into the technical and supply chain fields of the structured data containers corresponding to the second-level nodes, the third-level nodes, and the leaf nodes to obtain the subdivided structured data containers.

[0043] The second-level nodes, the third-level nodes, the leaf nodes, and their corresponding upstream and downstream edges are incorporated into the updated directed acyclic graph of the industrial chain to obtain the subdivided directed acyclic graph of the industrial chain.

[0044] Based on the primary key prefix of the intermediate results of the task, the intermediate results containing the second-level nodes, the third-level nodes, the leaf nodes, the corresponding upstream and downstream edges, the subdivided structured data containers, and the corresponding metadata are stored in a key-value pair set.

[0045] The local directed acyclic graph containing the second-level nodes, third-level nodes, leaf nodes and corresponding upstream and downstream edges of the subdivided industrial chain directed acyclic graph is visualized with the subdivided structured data container to change the subdivided chain building task from a pending state to a completed state, and to lock the second-level nodes, third-level nodes, leaf nodes and corresponding upstream and downstream edges.

[0046] Optionally, the step of using the supply chain collaboration system to perform supply chain labeling operations on the first-level nodes, the second-level nodes, the third-level nodes, and the leaf nodes to obtain supply chain labeling results includes:

[0047] After the scheduling agent detects that the subdivided chain building task is in a completed state, the scheduling agent is used to activate the chain labeling task so as to control the chain labeling task to be in a pending execution state.

[0048] When the industry chain labeling task is in a pending state, starting from the leaf node of the subdivided industry chain directed acyclic graph, the labeling function is called from bottom to top by the execution agent.

[0049] If it is a leaf node, the annotation function is applied to the subdivided structured data container corresponding to the leaf node to obtain the annotation results of the first and fourth dimensions;

[0050] If it is not a leaf node, the annotation function is applied to the union of the annotation results of the structured data container corresponding to the current node and the child nodes of the current node to obtain the second and fourth dimension annotation results;

[0051] The first four dimensions of annotation results and the second four dimensions of annotation results are used as the industry chain annotation results.

[0052] Optionally, based on the industry chain labeling results, the step of using the industry chain collaboration system to perform enterprise chain placement operations on the leaf nodes to obtain a chain placement relationship database and an industry chain panoramic map includes:

[0053] After the scheduling agent detects that the supply chain labeling task is in a completed state, it activates the enterprise's supply chain task to control the enterprise's supply chain task to be in a pending execution state.

[0054] Query the labeled leaf nodes in the industry chain labeling results;

[0055] When the enterprise's task is in a pending execution state, the execution agent runs the three-dimensional keyword engine in parallel on the labeled leaf nodes to generate technical keywords, product keywords and supply chain keywords.

[0056] The execution agent is used to filter enterprises from the enterprise database that simultaneously meet the technical keywords, product keywords, supply chain keywords, and have a region code equal to the current jurisdiction, thereby obtaining a set of matching enterprises;

[0057] Establish a bidirectional index between each labeled leaf node and the matching set of enterprises;

[0058] The enterprise-node bidirectional index is used to generate a chain relationship database;

[0059] The subdivided directed acyclic graph of the industry chain, which is equipped with matching enterprise sets, is marked as a panoramic map of the industry chain.

[0060] To address the aforementioned problems, this invention also provides an automatic chain establishment and termination system based on intelligent agent collaboration, the system comprising:

[0061] The system deployment module is used to deploy the supply chain collaboration system, which includes a set of intelligent agents, a set of task stages, a directed acyclic graph of the supply chain, and a set of key-value pairs.

[0062] The first-level chain building module is used to receive users' automatic chain building requests, and based on the automatic chain building requests, to perform first-level chain building operations on the directed acyclic graph of the industrial chain using the industrial chain collaboration system to obtain first-level nodes;

[0063] The subdivision chain building module is used to perform subdivision chain building operations on the first-level node using the industrial chain collaboration system to obtain second-level nodes, third-level nodes, and leaf nodes.

[0064] The supply chain labeling module is used to perform supply chain labeling operations on the first-level nodes, the second-level nodes, the third-level nodes, and the leaf nodes using the supply chain collaboration system to obtain supply chain labeling results;

[0065] The enterprise chain placement module is used to perform enterprise chain placement operations on the leaf nodes based on the industry chain labeling results and the industry chain collaboration system to obtain a chain placement relationship database and an industry chain panoramic map.

[0066] Compared to the problems described in the background art, this embodiment of the invention deploys a supply chain collaboration system, enabling the system state to dynamically evolve with user interaction and data updates, thus meeting dynamic requirements. Based on the automatic chain-building requirement, this embodiment utilizes the supply chain collaboration system to perform first-level chain-building operations on the directed acyclic graph of the supply chain. This allows for task parallelism combined with agent collaboration, reducing the time complexity from quadratic to linear logarithmic levels, thereby improving efficiency. Furthermore, this embodiment utilizes the supply chain collaboration system to perform subdivided chain-building operations on the first-level nodes, ensuring that the human-machine collaboration mechanism guarantees the graph meets user constraints and ensures controllability. New tasks can be added by expanding nodes and edges to maintain the acyclic property, achieving scalability. In terms of scalability, this invention utilizes the supply chain collaboration system to perform supply chain labeling operations on the primary, secondary, tertiary, and leaf nodes. This bottom-up labeling mechanism fills in the four dimensions of planning importance, core technologies, development priorities, and emerging technologies, forming labeling results that conform to regional policies and technological trends, thus improving policy consistency. Based on the supply chain labeling results, this invention uses the supply chain collaboration system to perform enterprise chain placement operations on the leaf nodes. This allows the three-dimensional keyword engine to integrate technology, product, and supply chain characteristics, achieving higher accuracy than traditional keyword fuzzy matching. It outputs a chain placement relationship database and a panoramic supply chain map, enabling dynamic updates and improved accuracy. Therefore, this invention can solve the problems of difficulty in dynamic chain building, lack of multi-stage collaboration, and low enterprise chain placement accuracy. Attached Figure Description

[0067] Figure 1 This is a flowchart illustrating an automatic chain establishment and termination method based on intelligent agent collaboration provided in an embodiment of the present invention.

[0068] Figure 2 This is an overall architecture diagram of an automatic chain establishment and termination method based on intelligent agent collaboration provided in an embodiment of the present invention;

[0069] Figure 3 This is a schematic diagram of a module for implementing the automatic chain establishment and termination system based on intelligent agent collaboration, provided in an embodiment of the present invention.

[0070] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0071] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0072] This application provides an automatic link establishment and termination method based on agent collaboration. The executing entity of this automatic link establishment and termination method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the automatic link establishment and termination method based on agent collaboration can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0073] Reference Figure 1 The diagram shown is a flowchart illustrating an automatic link establishment and termination method based on agent collaboration according to an embodiment of the present invention. In this embodiment, the automatic link establishment and termination method based on agent collaboration includes:

[0074] S1. Deploy the supply chain collaboration system, which includes a set of intelligent agents, a set of task stages, a directed acyclic graph of the supply chain, and a set of key-value pairs.

[0075] This invention deploys a supply chain collaboration system so that the system state can dynamically evolve with user interaction and data updates, thus meeting dynamic requirements.

[0076] In one embodiment of the present invention, the deployment of the supply chain collaboration system includes: initializing a set of intelligent agents, wherein the set of intelligent agents includes a planning intelligent agent, a scheduling intelligent agent, and an execution intelligent agent; initializing a set of task stages, wherein the set of task stages includes primary chain building tasks, subdivided chain building tasks, supply chain labeling tasks, and enterprise chain placement tasks; initializing a directed acyclic graph of the supply chain, wherein the directed acyclic graph of the supply chain includes a set of empty nodes and a set of empty edges; establishing a set of key-value pairs, wherein the set of key-value pairs includes a node context primary key prefix and a task intermediate result primary key prefix; and determining the supply chain collaboration system by the set of intelligent agents, the set of task stages, the directed acyclic graph of the supply chain, and the set of key-value pairs.

[0077] The planning agent parses user requirements, breaking them down into four stages: chain building / segmentation / labeling / chain placement, triggering user stage confirmation. The scheduling agent assigns tasks to the execution agent, monitors the status, and reports any anomalies to the planning agent. The execution agent calls tools such as getNodeName and buildMapTool to perform specific tasks and return structured results. The memory M_mem is defined as a key-value pair set: M_mem ={(k, v)|k∈K_key, v∈V}, where k = "context:" + v.id: stores the node context, and k = "task:" + This stores intermediate results of tasks. In other words, the key-value pair set includes two types of key-value pairs: those with "context:" as the primary key prefix for node context and those with "task:" as the primary key prefix for intermediate results of tasks. For all node contexts, the key is uniformly "context:" + node ID, and for all intermediate results of tasks, the key is uniformly "task:" + task name / task ID.

[0078] It should be noted that, It is a collection of intelligent agents, in which, PlannerAgent Scheduler Agent : Executor Agent : The set of task phases, where, Primary Chain Construction: : Refined Chain Construction ChainAnnotation: Enterprise Binding: G = (V, E) : Directed Acyclic Graph of the supply chain, where V: set of nodes, representing nodes in the supply chain, such as nodes in new energy vehicle manufacturing, battery systems, etc. Edge set, representing upstream and downstream relationships, satisfying And there is no ring.

[0079] In one embodiment of the present invention, after deploying the supply chain collaboration system, the method further includes: controlling the supply chain collaboration system to enter a state of waiting for user input.

[0080] Here, the user input refers to the industry planning question entered by the user, such as "the leading industry chain in YY City, XX Province". The industry chain collaboration system will then build the chain based on this industry planning question entered by the user.

[0081] S2. Receive the user's automatic chain building request. Based on the automatic chain building request, use the industrial chain collaboration system to perform a first-level chain building operation on the directed acyclic graph of the industrial chain to obtain a first-level node.

[0082] This invention, based on the automatic chain building requirement, utilizes the industrial chain collaboration system to perform a first-level chain building operation on the directed acyclic graph of the industrial chain. This allows for task parallelism combined with agent collaboration, reducing the time complexity from the quadratic level of manual work to the linear logarithmic level, thereby improving efficiency.

[0083] In one embodiment of the present invention, the step of performing a first-level chain building operation on the directed acyclic graph of the industrial chain based on the automatic chain building requirement and obtaining a first-level node using the industrial chain collaboration system includes: obtaining a set of agents and a set of task stages in the industrial chain collaboration system; obtaining a planning agent, a scheduling agent, and an execution agent in the set of agents; obtaining a first-level chain building task, a subdivided chain building task, an industrial chain labeling task, and an enterprise chain placement task in the set of task stages; and using the planning agent to parse the automatic chain building requirement to generate a task dependency graph composed of the first-level chain building task, the subdivided chain building task, the industrial chain labeling task, and the enterprise chain placement task. According to the topological order in the task dependency graph, the scheduling agent activates the first-level chain building task to keep it in a pending state. Simultaneously, the automatic chain building requirement generates a first-level chain building instruction for the task, which includes a region name and an industry name. After receiving the instruction, the execution agent invokes a mapping tool. The region name and industry name from the instruction are used as input parameters for the mapping tool. A joint query of the knowledge base and publicly available internet information is performed to output the first-level node corresponding to the input parameters using the mapping tool.

[0084] It should be noted that the process of using the planning agent to parse the automatic chain-building requirement to generate a task dependency graph consisting of the primary chain-building task, the subdivided chain-building task, the industry chain labeling task, and the enterprise chain-attaching task does not actually parse the specific descriptions within the automatic chain-building requirement. Rather, it means that the industry chain collaboration system receives the automatic chain-building requirement and determines whether this requirement calls for the industry chain collaboration system to build the chain. In short, the parsing of the automatic chain-building requirement here simply means that after receiving the automatic chain-building requirement, the industry chain collaboration system knows that it needs to start building the chain. Therefore, the automatic chain-building requirement plays a role in activating and awakening the industry chain collaboration system to begin operation.

[0085] Optionally, the step of using the planning agent to parse the automatic chain-building requirements to generate a task dependency graph consisting of the primary chain-building task, the subdivided chain-building task, the industry chain labeling task, and the enterprise chain placement task includes: a planning agent. Map user needs Q∈Q to task DAG = (T, D), where: D⊆T×T: Task dependency edges, defined as: D = {( , ), ( , ), ( , The condition D is a DAG (Directed Acyclic Graph), meaning that regardless of the user's requirements, the generated task dependency graph always consists of nodes. with edge {( , ), ( , ), ( , The directed acyclic graph (DAG) is constructed by} . The purpose of this configuration is to ensure that when receiving each user's request, it always originates from} . The task begins chain building. Further, according to the topological order in the task dependency graph, the scheduling agent activates the first-level chain building task to control it in a pending state. This includes: the scheduling agent... Sort by topology Topo(D) = [ Activate the tasks one by one. Here's something to note: ,initial pending, the rest are pending, when completed, and Completed, then activated. "Pending" indicates a pending execution status, while "completed" indicates a completed status. Indicates the running status. Indicating a failure status, further, while activating the primary chain-building task, the primary chain-building instruction for the primary chain-building task is generated using the automatic chain-building requirement. This primary chain-building instruction includes a region name and an industry name. This primary chain-building instruction is also the command space CMD, where CMD = { = (agent, task, params, callback)|agent∈A, task∈T, params∈P, callback∈F}, where P is the parameter space, F is the set of callback functions, and params is used to store region names and industry names, for example... = ( , , {region:"XX City", industry: "new energy vehicle"}, buildMapTool), buildMapTool is the local mapping tool for the executing agent. It can be invoked and run via this CMD command. Further, the region name and industry name in the first-level chain-building command are used as input parameters for the mapping tool, including: the executing agent only retrieves params={region:"XX City", industry: "new energy vehicle"} from the CMD and inputs it into buildMapTool. Further, a joint query of the knowledge base and publicly available internet information is performed to output the first-level node corresponding to the input parameters through the mapping tool, including: the input to buildMapTool is {region, The internal implementation of buildMapTool is as follows: Dual-source query: knowledge base (patents, policies, research institutions) and the Internet: industry reports, regional planning white papers. The action of querying publicly available information from the knowledge base and the Internet is the input step of buildMapTool itself. The output simultaneously provides two things: graph structure: the set of first-level nodes and upstream and downstream edges; node content: the values ​​of fields such as attr and policy for each node. attr is used to fill in industry attributes, and policy is used to fill in regional constraint information. The contents of the attr and policy fields are the JSON segments returned by buildMapTool immediately after querying the knowledge base and the Internet. The output of buildMapTool also includes metadata, such as "result": {"graph": { "nodes": [...], "edges": [...]},"metadata": { "source": ["patent", "policy"], "confidence":0.95}}. Here, source and confidence in metadata are the metadata. graph represents the graph structure, and the metadata is the JSON field returned by the graphing tool, containing the data source identifier source and confidence score confidence.

[0086] In one embodiment of the present invention, after performing a first-level chain building operation on the directed acyclic graph of the industrial chain based on the automatic chain building requirement and obtaining the first-level nodes using the industrial chain collaboration system, the method further includes: obtaining the upstream and downstream edges, metadata, industry attributes, and constraint information of the first-level nodes returned by the graph building tool; writing the industry attributes and constraint information into the attribute fields and constraint fields of the structured data containers of each first-level node to obtain an updated structured data container; and incorporating the first-level nodes and the upstream and downstream edges into the directed acyclic graph of the industrial chain to obtain an updated directed acyclic graph of the industrial chain. An acyclic graph is constructed. Based on the primary key prefix of the intermediate results of the task, the intermediate results containing the first-level nodes, the upstream and downstream edges, the updated structured data container, and the metadata are stored in a key-value pair set. The local directed acyclic graph containing the first-level nodes and the upstream and downstream edges of the updated industry chain directed acyclic graph and the updated structured data container are visualized to obtain a visualization interface. It is detected whether the user has triggered a confirmation operation on the visualization interface. After the confirmation operation is triggered, the pending state of the first-level chain building task is changed to the completed state, and the first-level nodes and the upstream and downstream edges are locked.

[0087] The structured data container mentioned above is a structured data container (JSON object) that the system immediately creates for each node, that is... , :node The context is defined as structured data. Each field is a JSON object.

[0088] Optionally, the step of incorporating the first-level node and the upstream and downstream edges into the directed acyclic graph of the industrial chain to obtain the updated directed acyclic graph of the industrial chain includes: since the initial directed acyclic graph of the industrial chain is empty, the incorporation here is actually assembling the first-level node and the upstream and downstream edges into a directed acyclic graph. Further, the updated directed acyclic graph of the industrial chain containing the first-level node and the upstream and downstream edges, along with the updated structured data container, is visualized, including: during visualization, returning getNodeName(v) = (v.name, C(v)). The displayed content includes the attr / policy in C(v). That is, when the system visualizes, it directly calls the getNodeName function to render the returned node name and context (including fields such as attr / policy) to the user. Furthermore, detecting whether the user triggers a confirmation operation on the visualization interface includes: the aforementioned visualization is for the user; after viewing it, the user will perform some operations. User operations u∈U act on the current graph G, where U is the user interaction space, defined as the operation set U={add, delete, modify, reconnect, confirm}, and the operation functions are defined as follows: ; ;modify ; When the user executes confirm The scheduling agent triggers DAG reconstruction: if and Then reset And regenerate the subgraph, if policy The inheritance strategy is then passed to the child node set (children). ): children policy policy Here, DAG reconstruction refers to the recalculation of a local subgraph of the directed acyclic graph G of the supply chain, not the reconstruction of the task-dependent graph D.

[0089] It should be noted that the purpose of storing intermediate results in the key-value pair set is: if the user deletes / modifies a node, the system will retrieve the result from M_mem["task: Rollback to the current snapshot (the current snapshot refers to the previous nodes, edges, ...). (and metadata), when the pending status of the first-level chain-building task is changed to the completed status after user confirmation, this M_mem["task: The key is used for rollback in cases of exception recovery and auditing. In other words, intermediate results are stored in a key-value pair set to preserve data for user editing and status auditing. Modification operations introduced through human-computer interaction cause local subgraphs of the task DAG. If a value is marked as dirty, the scheduling agent resets its state to pending to ensure system consistency.

[0090] S3. Utilize the aforementioned industry chain collaboration system to perform subdivision chain building operations on the primary nodes to obtain secondary nodes, tertiary nodes, and leaf nodes.

[0091] This invention utilizes the industry chain collaboration system to perform subdivided chain building operations on the primary nodes, so that the human-machine collaboration mechanism ensures that the graph meets user constraints and ensures controllability. New tasks can be added by expanding nodes and edges to maintain acyclic properties, thus achieving scalability.

[0092] In one embodiment of the present invention, the step of using the supply chain collaboration system to perform subdivided chain building operations on the first-level nodes to obtain second-level nodes, third-level nodes, and leaf nodes includes: after the scheduling agent detects that the first-level chain building task is in a completed state, activating the subdivided chain building task using the scheduling agent to control the subdivided chain building task to be in a pending execution state; using the execution agent to traverse the locked first-level nodes in the updated directed acyclic graph of the supply chain to call the updated structured data container corresponding to the first-level node; inheriting the constraint information in the updated structured data container; using the first-level node as the parent node, using the execution agent to call the graph building tool to query business information, patent clustering results, and supply chain data; and generating the business information, patent clustering results, and supply chain data as second-level nodes, third-level nodes, and leaf nodes.

[0093] Optionally, inheriting the constraint information in the updated structured data container includes: defining a node context inheritance function I:V→P(V)×C, used for subdividing chain building: I(v_parent)=(children(v_parent), C(v_parent)) means copying the entire context of the parent node to the child node, that is, copying the content of the first-level node to the child node of the first-level node. Further, taking the first-level node as the parent node, the execution agent calls the graph building tool to query business information, patent clustering results and supply chain data, including: for each parent node v_parent, calling buildMapTool again for deep query: business information: registered capital, survival years; patent clustering: technology hotspot map; supply chain data: cross-validation. Further, generating the business information, patent clustering results and supply chain data into second-level nodes, third-level nodes and leaf nodes includes: buildMapTool takes the queried business information, patent clustering and supply chain data as input features and outputs the corresponding second-level, third-level and leaf nodes (including node name, upstream and downstream edges and node context fields), for example, the second-level node "battery system" and the third-level node "cathode material / separator / electrolyte".

[0094] In one embodiment of the present invention, after performing subdivision and chain-building operations on the first-level nodes using the supply chain collaboration system to obtain second-level nodes, third-level nodes, and leaf nodes, the method further includes: writing business information, patent clustering results, and supply chain data into the technical and supply chain fields of the structured data containers corresponding to the second-level nodes, third-level nodes, and leaf nodes to obtain subdivided structured data containers; merging the second-level nodes, third-level nodes, leaf nodes, and their corresponding upstream and downstream edges into the updated directed acyclic graph of the supply chain to obtain the subdivided directed acyclic graph of the supply chain; based on the task... The intermediate results, including the secondary nodes, tertiary nodes, leaf nodes, corresponding upstream and downstream edges, the subdivided structured data container, and corresponding metadata, are stored in a key-value pair set. The subdivided industry chain directed acyclic graph containing the secondary nodes, tertiary nodes, leaf nodes, and corresponding upstream and downstream edges, along with the subdivided structured data container, are visualized to change the subdivided chain building task from a pending state to a completed state, and to lock the secondary nodes, tertiary nodes, leaf nodes, and corresponding upstream and downstream edges.

[0095] Optionally, this section will explain the following content after using the aforementioned industry chain collaboration system to perform subdivided chain-building operations on the first-level nodes to obtain second-level nodes, third-level nodes, and leaf nodes. Since the execution process is similar to the previous steps, each sentence will not be explained here: Generating a set of second-level nodes. Third-level node set and leaf node set Write layer by layer The tech and supply fields; , and corresponding edges join in ,renew Visualize the current full image and wait for user interaction; only after the user confirms... Change to "completed", lock all newly added nodes and edges; and mark them with "task: "Store the intermediate results into a key-value pair set; visualize the local graph containing the second-level nodes, third-level nodes, and leaf nodes, as well as the structured data container of each node, and detect whether the user has triggered a confirmation operation; after the confirmation operation is triggered, set the status of the subdivided chain building task to the completed state."

[0096] S4. Use the supply chain collaboration system to perform supply chain labeling operations on the first-level nodes, the second-level nodes, the third-level nodes and the leaf nodes to obtain supply chain labeling results.

[0097] This invention utilizes the industry chain collaboration system to perform industry chain labeling operations on the first-level nodes, second-level nodes, third-level nodes, and leaf nodes. This bottom-up labeling mechanism fills in the four dimensions of planning importance, core technologies, development priorities, and emerging technologies, forming labeling results that conform to regional policies and technology trends, thereby improving policy consistency.

[0098] Among them, the importance of planning refers to the priority of support for this node in regional industrial policies; core technology refers to the maturity of key technologies involved in this node and patent layout, etc.; development focus refers to the development tasks clearly put forward in the plans of relevant departments or industries; and emerging technology refers to the cutting-edge or alternative technology directions involved in this node.

[0099] In one embodiment of the present invention, the step of using the supply chain collaboration system to perform supply chain labeling operations on the first-level nodes, the second-level nodes, the third-level nodes, and the leaf nodes to obtain supply chain labeling results includes: after the scheduling agent detects that the subdivided supply chain building task is in a completed state, the scheduling agent activates the supply chain labeling task to control the supply chain labeling task to be in a pending execution state; when the supply chain labeling task is in a pending execution state, starting from the leaf nodes in the subdivided directed acyclic graph of the supply chain, the execution agent calls the labeling function from bottom to top; if it is a leaf node, the labeling function is applied to the subdivided structured data container corresponding to the leaf node to obtain the first four-dimensional labeling result; if it is not a leaf node, the labeling function is applied to the union of the labeling results of the structured data container corresponding to the current node and the child nodes of the current node to obtain the second four-dimensional labeling result; the first four-dimensional labeling result and the second four-dimensional labeling result are used as the supply chain labeling result.

[0100] Optionally, this section still only explains the overall interpretation of the supply chain labeling results obtained by performing supply chain labeling operations on the first-level nodes, second-level nodes, third-level nodes, and leaf nodes using the aforementioned supply chain collaboration system: The scheduling agent detected... Completed, activated Set to pending; from the leaf node set Begin by calling Annotate(v) from the bottom up: if -label has a direct effect Otherwise, f-label acts on children (v) Annotate ( Each node is labeled with four dimensions: planning importance, core technologies (technology maturity and patent information), key development documents, and emerging technologies; the labeling results are written back to C(v) and M-mem["context: " + v. id] is updated; the labeled map is visualized and awaits user review or modification; after user confirmation, The annotation task is now complete, the annotation results are locked. Defined from leaf nodes out-degree Mapping back to the root node:

[0101] Annotate(v) =

[0102] f_label(C(v)), if v ∈ L

[0103] ,otherwise

[0104] Here, f_label is the labeling function, which outputs four dimensions: planning importance, core technologies (including TRL and patents), development priorities, and emerging technologies.

[0105] S5. Based on the industry chain labeling results, the industry chain collaboration system is used to perform enterprise chain placement operations on the leaf nodes to obtain a chain placement relationship database and an industry chain panoramic map.

[0106] This invention, based on the industry chain annotation results, utilizes the industry chain collaboration system to perform enterprise chain placement operations on the leaf nodes, enabling the 3D keyword engine to integrate technology, product, and supply chain features, achieving higher accuracy than traditional keyword fuzzy matching. It outputs a chain placement relationship database and an industry chain panoramic map, realizing dynamic updates and improved accuracy.

[0107] In one embodiment of the present invention, the step of performing enterprise chain placement operations on the leaf nodes based on the industry chain labeling results and using the industry chain collaboration system to obtain a chain placement relationship database and an industry chain panoramic map includes: after the scheduling agent detects that the industry chain labeling task is in a completed state, activating the enterprise chain placement task using the scheduling agent to control the enterprise chain placement task to be in a pending execution state; querying the labeled leaf nodes in the industry chain labeling results; when the enterprise chain placement task is in a pending execution state, using the execution agent to run a three-dimensional keyword engine in parallel on the labeled leaf nodes to generate technical keywords, product keywords, and supply chain keywords; using the execution agent to filter enterprises from the enterprise database that simultaneously satisfy the technical keywords, product keywords, and supply chain keywords and whose region code is equal to the current jurisdiction to obtain a matching enterprise set; establishing an enterprise-node bidirectional index for each labeled leaf node and the matching enterprise set; generating a chain placement relationship database using the enterprise-node bidirectional index; and marking the subdivided industry chain directed acyclic graph with matching enterprise sets as an industry chain panoramic map.

[0108] Optionally, this section only explains the overall process: After the scheduling agent detects that the supply chain labeling task is completed, it activates the enterprise's supply chain assignment task and sets it to pending execution; the execution agent runs a three-dimensional keyword engine in parallel on the labeled leaf nodes. This three-dimensional keyword engine specifically includes: generating technical keywords by adding the first three digits of the patent IPC to the core technology verb, generating product keywords by adding the product standard name to the application scenario, and generating supply chain keywords by adding the raw material HS code to the process characteristics. Then, it selects enterprises from the enterprise database that simultaneously meet the three types of keywords and whose regional code is equal to the current jurisdiction to form a matching set. A bidirectional index of "enterprise-node" is established for each leaf node and written into the supply chain relationship database. The directed acyclic graph of the complete supply chain with attached enterprises is marked as a panoramic supply chain map, with "task: "Store the key into a key-value pair set, and finally set the enterprise's chain task status to complete and output the chain relationship database and the industry chain panorama map. The enterprise is attached under the labeled leaf node. It should also be noted that S5 only uses the leaf node. The second-level and third-level nodes are used by the S4 step for users to view."

[0109] See Figure 2 The diagram shown is an overall architecture diagram of an automatic chain establishment and termination method based on agent collaboration provided in an embodiment of the present invention. Figure 2 In the process, the user inputs the chain-building requirements (e.g., "Building a leading industrial chain in YY City, XX Province"); the planning agent breaks down the task into four stages: First, a Level 1 chain-building task is initiated, where the agent generates Level 1 industrial chain nodes through combined queries; after the user edits and confirms the nodes, a more detailed chain-building task is triggered, where the agent generates Level 2 to N industrial chain nodes through combined queries; after the user edits and confirms the nodes, an industrial chain labeling task is triggered, where the labeling stage traces back from the leaf nodes, filling in information across four dimensions: planning importance, core technologies, development priorities, and emerging technologies; after the user confirms the industrial chain labeling information, an enterprise chain implementation task is triggered, where the implementation stage generates three sets of key data: technology dimension, product dimension, and supply chain dimension. Keywords are used to filter enterprises and attach them to the corresponding leaf nodes. After the user confirms the enterprise attachment results, a supply chain map is generated. The stage transition confirmation in the map refers to the user completing the current stage (such as first-level chain building, sub-chain building, supply chain labeling, or enterprise chain placement) and confirming that there are no errors through the confirm operation in the user operation set U. The system receives the confirmation signal, updates the current task status from pending to completed, and activates the next stage task. This confirmation operation is a necessary condition for the process to proceed and mainly includes a series of operations that require user confirmation, such as user editing and confirmation of nodes, user confirmation of supply chain labeling information, and user confirmation of enterprise attachment results.

[0110] Compared to the problems described in the background art, this embodiment of the invention deploys a supply chain collaboration system, enabling the system state to dynamically evolve with user interaction and data updates, thus meeting dynamic requirements. Based on the automatic chain-building requirement, this embodiment utilizes the supply chain collaboration system to perform first-level chain-building operations on the directed acyclic graph of the supply chain. This allows for task parallelism combined with agent collaboration, reducing the time complexity from quadratic to linear logarithmic levels, thereby improving efficiency. Furthermore, this embodiment utilizes the supply chain collaboration system to perform subdivided chain-building operations on the first-level nodes, ensuring that the human-machine collaboration mechanism guarantees the graph meets user constraints and ensures controllability. New tasks can be added by expanding nodes and edges to maintain the acyclic property, achieving scalability. In terms of scalability, this invention utilizes the supply chain collaboration system to perform supply chain labeling operations on the primary, secondary, tertiary, and leaf nodes. This bottom-up labeling mechanism fills in the four dimensions of planning importance, core technologies, development priorities, and emerging technologies, forming labeling results that conform to regional policies and technological trends, thus improving policy consistency. Based on the supply chain labeling results, this invention uses the supply chain collaboration system to perform enterprise chain placement operations on the leaf nodes. This allows the three-dimensional keyword engine to integrate technology, product, and supply chain characteristics, achieving higher accuracy than traditional keyword fuzzy matching. It outputs a chain placement relationship database and a panoramic supply chain map, enabling dynamic updates and improved accuracy. Therefore, this invention can solve the problems of difficulty in dynamic chain building, lack of multi-stage collaboration, and low enterprise chain placement accuracy.

[0111] like Figure 3 The diagram shown is a functional block diagram of an automatic chain establishment and termination system based on intelligent agent collaboration according to the present invention.

[0112] The automatic chain establishment and connection system 300 based on intelligent agent collaboration described in this invention can be installed in an electronic device. Depending on the functions implemented, the automatic chain establishment and connection system based on intelligent agent collaboration may include a system deployment module 301, a primary chain establishment module 302, a subdivided chain establishment module 303, a supply chain labeling module 304, and an enterprise connection module 305. The modules described in this invention can also be called units, referring to a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, stored in the memory of the electronic device.

[0113] In this embodiment of the invention, the functions of each module / unit are as follows:

[0114] The system deployment module 301 is used to deploy the industrial chain collaboration system, which includes a set of intelligent agents, a set of task stages, a directed acyclic graph of the industrial chain, and a set of key-value pairs.

[0115] The first-level chain building module 302 is used to receive the user's automatic chain building request, and based on the automatic chain building request, use the industrial chain collaboration system to perform a first-level chain building operation on the directed acyclic graph of the industrial chain to obtain a first-level node;

[0116] The subdivided chain building module 303 is used to perform subdivided chain building operations on the first-level node using the industrial chain collaboration system to obtain second-level nodes, third-level nodes and leaf nodes.

[0117] The supply chain labeling module 304 is used to perform supply chain labeling operations on the first-level nodes, the second-level nodes, the third-level nodes and the leaf nodes using the supply chain collaboration system to obtain supply chain labeling results;

[0118] The enterprise chain placement module 305 is used to perform enterprise chain placement operations on the leaf nodes based on the industry chain labeling results and the industry chain collaboration system to obtain a chain placement relationship database and an industry chain panoramic map.

[0119] In detail, the modules in the automatic link establishment and connection termination system 300 based on intelligent agent collaboration described in this embodiment of the invention employ the same methods as described above. Figure 1 The method used here is the same as the automatic chain establishment and termination method based on intelligent agent collaboration, and can produce the same technical effect, so it will not be elaborated here.

[0120] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0121] Finally, it should be noted that in the above embodiments, each embodiment can be combined with each other or independent. Deleting any one of them will not affect the technical implementation of other embodiments. The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An automatic chain establishment and termination method based on agent collaboration, characterized in that, The method includes: Deploy a supply chain collaboration system, which includes a set of intelligent agents, a set of task stages, a directed acyclic graph of the supply chain, and a set of key-value pairs; Receive user's automatic chain building request, and based on the automatic chain building request, use the industrial chain collaboration system to perform a first-level chain building operation on the directed acyclic graph of the industrial chain to obtain a first-level node. Specifically, this includes: obtaining the set of intelligent agents and the set of task stages in the industrial chain collaboration system; obtaining the planning intelligent agent, scheduling intelligent agent and execution intelligent agent in the set of intelligent agents; and obtaining the first-level chain building task, subdivided chain building task, industrial chain labeling task and enterprise chain landing task in the set of task stages. The planning agent is used to analyze the automatic chain building requirements to generate a task dependency graph consisting of the first-level chain building task, the subdivided chain building task, the industry chain labeling task, and the enterprise chain placement task. According to the topological order in the task dependency graph, the scheduling agent is used to activate the first-level chain building task so as to control the first-level chain building task to be in a pending execution state. While activating the primary chain building task, the primary chain building instruction of the primary chain building task is generated using the automatic chain building requirement, wherein the primary chain building instruction includes the region name and the industry name; After receiving the first-level chain-building instruction, the execution agent invokes the graph-building tool. The region name and industry name in the first-level chain building instruction are used as input parameters for the mapping tool; By jointly querying knowledge bases and publicly available information on the Internet, the first-level nodes corresponding to the input parameters are output through the graphing tool. The supply chain collaboration system is used to perform subdivision chain building operations on the first-level nodes to obtain second-level nodes, third-level nodes, and leaf nodes. The supply chain collaboration system is used to perform supply chain labeling operations on the first-level nodes, the second-level nodes, the third-level nodes, and the leaf nodes to obtain supply chain labeling results; Based on the industry chain labeling results, the industry chain collaboration system is used to perform enterprise chain placement operations on the leaf nodes to obtain a chain placement relationship database and an industry chain panoramic map.

2. The automatic link establishment and termination method based on agent collaboration as described in claim 1, characterized in that, The deployment of the industry chain collaboration system includes: Initialize a set of intelligent agents, wherein the set of intelligent agents includes a planning intelligent agent, a scheduling intelligent agent, and an execution intelligent agent; An initial task phase set is established, which includes primary chain building tasks, subdivided chain building tasks, industry chain labeling tasks, and enterprise chain placement tasks. Initialize a directed acyclic graph of supply chains, wherein the directed acyclic graph of supply chains includes a set of empty nodes and a set of empty edges; Establish a set of key-value pairs, wherein the set of key-value pairs includes a node context primary key prefix and a task intermediate result primary key prefix; The supply chain collaboration system is determined by the set of intelligent agents, the set of task stages, the directed acyclic graph of the supply chain, and the set of key-value pairs.

3. The automatic link establishment and termination method based on agent collaboration as described in claim 1, characterized in that, After deploying the supply chain collaboration system, the following is also included: The system is controlled to enter a state where it awaits user input.

4. The automatic link establishment and termination method based on agent collaboration as described in claim 1, characterized in that, Based on the automatic chain building requirement, after performing a first-level chain building operation on the directed acyclic graph of the industrial chain using the industrial chain collaboration system to obtain the first-level nodes, the process further includes: Obtain the upstream and downstream edges, metadata, industry attributes, and constraint information of the first-level nodes connected by the mapping tool; The industry attributes and the constraint information are written into the attribute fields and constraint fields of the structured data container of each first-level node in the first-level node, respectively, to obtain the updated structured data container; The first-level node and the upstream and downstream edges are incorporated into the directed acyclic graph of the industrial chain to obtain the updated directed acyclic graph of the industrial chain. Based on the primary key prefix of the intermediate results of the task, the intermediate results containing the first-level node, the upstream and downstream edges, the updated structured data container and the metadata are stored in a key-value pair set; The updated directed acyclic graph of the industrial chain, which includes the first-level node and the upstream and downstream edges, is visualized along with the updated structured data container to obtain a visualization interface. Detect whether the user has triggered a confirmation operation on the visual interface; After the confirmation operation is triggered, the pending state of the first-level chain building task is changed to the completed state, and the first-level node and the upstream and downstream edges are locked.

5. The automatic link establishment and termination method based on agent collaboration as described in claim 1, characterized in that, The process of using the supply chain collaboration system to perform subdivided chain-building operations on the primary nodes to obtain secondary nodes, tertiary nodes, and leaf nodes includes: After the scheduling agent detects that the primary chain building task is in a completed state, it activates the sub-chain building task to control the sub-chain building task to be in a pending execution state. The execution agent traverses the locked first-level nodes in the updated directed acyclic graph of the industry chain to call the updated structured data container corresponding to the first-level node. Inherit the constraint information from the updated structured data container; Using the first-level node as the parent node, the execution agent calls the mapping tool to query business information, patent clustering results, and supply chain data; The business registration information, the patent clustering results, and the supply chain data are generated as secondary nodes, tertiary nodes, and leaf nodes.

6. The automatic link establishment and termination method based on agent collaboration as described in claim 1, characterized in that, After using the supply chain collaboration system to perform subdivision and chain-building operations on the primary nodes to obtain secondary nodes, tertiary nodes, and leaf nodes, the process further includes: Business information, patent clustering results, and supply chain data are written into the technical and supply chain fields of the structured data containers corresponding to the second-level nodes, the third-level nodes, and the leaf nodes to obtain the subdivided structured data containers. The second-level nodes, the third-level nodes, the leaf nodes, and their corresponding upstream and downstream edges are incorporated into the updated directed acyclic graph of the industrial chain to obtain the subdivided directed acyclic graph of the industrial chain. Based on the primary key prefix of the intermediate results of the task, the intermediate results containing the second-level nodes, the third-level nodes, the leaf nodes, the corresponding upstream and downstream edges, the subdivided structured data containers, and the corresponding metadata are stored in a key-value pair set. The local directed acyclic graph containing the second-level nodes, third-level nodes, leaf nodes, and corresponding upstream and downstream edges of the subdivided industrial chain directed acyclic graph is visualized with the subdivided structured data container to transform the subdivided chain building task from a pending state to a completed state, and to lock the second-level nodes, third-level nodes, leaf nodes, and corresponding upstream and downstream edges.

7. The automatic link establishment and termination method based on agent collaboration as described in claim 1, characterized in that, The process of using the supply chain collaboration system to perform supply chain labeling operations on the first-level nodes, second-level nodes, third-level nodes, and leaf nodes to obtain supply chain labeling results includes: After the scheduling agent detects that the subdivided chain building task is in a completed state, the scheduling agent is used to activate the chain labeling task so as to control the chain labeling task to be in a pending execution state. When the industry chain labeling task is in a pending state, starting from the leaf node of the subdivided industry chain directed acyclic graph, the labeling function is called from bottom to top by the execution agent. If it is a leaf node, the annotation function is applied to the subdivided structured data container corresponding to the leaf node to obtain the annotation results of the first and fourth dimensions; If it is not a leaf node, the annotation function is applied to the union of the annotation results of the structured data container corresponding to the current node and the child nodes of the current node to obtain the second and fourth dimension annotation results; The first four dimensions of annotation results and the second four dimensions of annotation results are used as the industry chain annotation results.

8. The automatic link establishment and termination method based on agent collaboration as described in claim 1, characterized in that, Based on the industry chain annotation results, the industry chain collaboration system is used to perform enterprise chain placement operations on the leaf nodes to obtain a chain placement relationship database and an industry chain panoramic map, including: After the scheduling agent detects that the supply chain labeling task is in a completed state, it activates the enterprise's supply chain task to control the enterprise's supply chain task to be in a pending execution state. Query the labeled leaf nodes in the industry chain labeling results; When the enterprise's task is in a pending execution state, the execution agent runs the three-dimensional keyword engine in parallel on the labeled leaf nodes to generate technical keywords, product keywords and supply chain keywords. The execution agent is used to filter enterprises from the enterprise database that simultaneously meet the technical keywords, product keywords, supply chain keywords, and have a region code equal to the current jurisdiction, thereby obtaining a set of matching enterprises; Establish a bidirectional index between each labeled leaf node and the matching set of enterprises; The enterprise-node bidirectional index is used to generate a chain relationship database; The subdivided directed acyclic graph of the industry chain, which is equipped with matching enterprise sets, is marked as a panoramic map of the industry chain.

9. An automatic chain establishment and termination system based on agent collaboration, characterized in that, The system includes: The system deployment module is used to deploy the supply chain collaboration system, which includes a set of intelligent agents, a set of task stages, a directed acyclic graph of the supply chain, and a set of key-value pairs. The first-level chain building module is used to receive users' automatic chain building requests. Based on the automatic chain building requests, it uses the industrial chain collaboration system to perform first-level chain building operations on the directed acyclic graph of the industrial chain to obtain first-level nodes. Specifically, it includes: obtaining the set of intelligent agents and the set of task stages in the industrial chain collaboration system; obtaining the planning intelligent agent, scheduling intelligent agent and execution intelligent agent in the set of intelligent agents; and obtaining the first-level chain building tasks, subdivided chain building tasks, industrial chain labeling tasks and enterprise chain placement tasks in the set of task stages. The planning agent is used to analyze the automatic chain building requirements to generate a task dependency graph consisting of the first-level chain building task, the subdivided chain building task, the industry chain labeling task, and the enterprise chain placement task. According to the topological order in the task dependency graph, the scheduling agent is used to activate the first-level chain building task so as to control the first-level chain building task to be in a pending execution state. While activating the primary chain building task, the primary chain building instruction of the primary chain building task is generated using the automatic chain building requirement, wherein the primary chain building instruction includes the region name and the industry name; After receiving the first-level chain-building instruction, the execution agent invokes the graph-building tool. The region name and industry name in the first-level chain building instruction are used as input parameters for the mapping tool; By jointly querying knowledge bases and publicly available information on the Internet, the first-level nodes corresponding to the input parameters are output through the graphing tool. The subdivision chain building module is used to perform subdivision chain building operations on the first-level node using the industrial chain collaboration system to obtain second-level nodes, third-level nodes, and leaf nodes. The supply chain labeling module is used to perform supply chain labeling operations on the first-level nodes, the second-level nodes, the third-level nodes, and the leaf nodes using the supply chain collaboration system to obtain supply chain labeling results; The enterprise chain placement module is used to perform enterprise chain placement operations on the leaf nodes based on the industry chain labeling results and the industry chain collaboration system to obtain a chain placement relationship database and an industry chain panoramic map.

Images