A method for identifying needs and configuring services of a technology transfer service

Abstract

The application discloses a kind of technical transfer service demand identification and service configuration method, comprising: based on the preset knowledge graph construction mode, the overall technology implementation process of target enterprise is constructed into enterprise technology knowledge graph, and multiple candidate new technologies are respectively constructed into candidate technology knowledge graph;Candidate technology knowledge graph and enterprise technology knowledge graph are graph matched, and the blocking node and blocking edge that cannot be matched are identified;Further determine the minimum graph editing operation quantity required for eliminating all blocking nodes and blocking edges to enterprise technology knowledge graph, operation includes the replacement and insertion of node and the replacement and insertion of edge;Finally, according to the minimum graph editing operation quantity, candidate technology knowledge graph is sorted, and technology transfer matching and service are carried out according to the sorting result.The application solves the problem that the existing technology cannot quantitatively evaluate the compatibility of new technology and the existing system of enterprise and the difficulty of modification.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and more specifically, to a method for demand identification and service configuration of technology transfer services. Background Technology

[0002] In the current process of technology transfer and enterprise upgrading and transformation, accurate matching of technology supply and demand is a core element. Typically, technology transfer services aim to help companies introduce advanced external technologies to improve their production efficiency or product quality. To achieve this goal, service providers often need to retrieve potential candidate technologies from massive technology databases and recommend them to target companies with relevant needs.

[0003] However, existing technology transfer matching methods mostly rely on shallow feature matching based on keywords or text classification. While this approach can filter out technologies that are functionally relevant, it often overlooks compatibility and adaptation costs during technology implementation. A company's technology implementation environment is typically a complex system, including specific processes, existing hardware, and fixed process parameter constraints. A theoretically advanced technology, if significantly different from a company's existing production system, may require substantial modification costs, such as large-scale equipment replacement or restructuring of core processes. This reduces the cost-effectiveness of technology introduction and may even lead to project failure. Currently, the industry lacks a technical analysis tool that can delve into the granularity of processes and equipment to quantitatively assess the difficulty of modifying a new technology and formulate specific implementation plans accordingly. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method for demand identification and service configuration of technology transfer services, so as to solve the problems mentioned in the background art. To achieve the above objectives, the present invention adopts the following technical solution: A method for demand identification and service configuration of technology transfer services includes the following steps: Based on a pre-defined knowledge graph construction method, the overall technology implementation process of the target enterprise is constructed into an enterprise technology knowledge graph; and based on the same knowledge graph construction method as the enterprise technology knowledge graph, multiple candidate new technologies are constructed into candidate technology knowledge graphs respectively. Graph matching is performed between the candidate technology knowledge graph and the enterprise technology knowledge graph. When a node in the candidate technology knowledge graph cannot be matched with the enterprise technology knowledge graph, the corresponding technology node is marked as a blocked node. When a conditional edge in the candidate technology knowledge graph cannot be matched with the enterprise technology knowledge graph, the corresponding conditional edge is marked as a blocked edge. The minimum number of graph editing operations required to eliminate all the blocked nodes and blocked edges in the enterprise technical knowledge graph is determined, and the graph editing operations include node replacement, node insertion, edge replacement, and edge insertion. Multiple candidate technology knowledge graphs are sorted, with the candidate technology knowledge graph having fewer minimum graph editing operations ranked higher, and the corresponding candidate technology technology transfer matching and services are performed based on the ranking.

[0005] Preferably, the preset knowledge graph construction method is as follows: Each step, each piece of hardware, and each process parameter in the technology implementation process is defined as a physical node. Define the logical flow relationship between processes as the timing edge, define the subordinate relationship between processes and hardware devices as the composition edge, and define the constraint relationship between process parameters and processes as the attribute edge. By extracting entity nouns and verbs from the technical documents of the target enterprise, and mapping them to entity nodes, temporal edges, composition edges and attribute edges, the enterprise's technical knowledge graph is generated. By extracting entity nouns and verbs from the technical specifications of the candidate new technologies and mapping them to entity nodes, temporal edges, composition edges, and attribute edges, a knowledge graph of the candidate technologies is generated.

[0006] Preferably, the steps for graph matching between the candidate technology knowledge graph and the enterprise technology knowledge graph specifically include: Each node in the candidate technology knowledge graph is transformed into a candidate node semantic vector using a pre-trained natural language processing model, and each node in the enterprise technology knowledge graph is transformed into an enterprise node semantic vector. Calculate the cosine similarity between the semantic vector of each candidate node and the semantic vector of each enterprise node; When the cosine similarity between a node in the candidate technology knowledge graph and any node in the enterprise technology knowledge graph is less than a preset similarity threshold, the node is determined to be unmatchable and is marked as the blocking node.

[0007] Preferably, the step of marking blocking edges specifically includes: In the candidate technology knowledge graph, the connection edge between two matched nodes is identified as the edge to be verified. Search for two enterprise nodes in the enterprise technology knowledge graph that correspond to the two matched nodes; Determine whether there is a connection edge between the two enterprise nodes that matches the attribute type of the edge to be verified. If there is no connecting edge between the two enterprise nodes that matches the attribute type of the edge to be verified, or if the direction of the connecting edge between the two enterprise nodes is opposite to the direction of the edge to be verified, then the edge to be verified is marked as the blocking edge.

[0008] Preferably, the step of determining the minimum number of image editing operations specifically includes: Construct a state space search tree, taking the initial state of the enterprise technology knowledge graph as the root node and the graph state after eliminating one of the blocking nodes or one of the blocking edges as the child nodes. Define the path weights of the four operations—node replacement, node insertion, edge replacement, and edge insertion—as 1. The A* search algorithm is used to find the shortest path from the initial state to the target state that is isomorphic to the candidate technology knowledge graph in the state space search tree. The total number of node replacements, node insertions, edge replacements, and edge insertions along the shortest path is counted, and this total number is determined as the minimum number of graph editing operations.

[0009] Preferably, after the step of ranking multiple candidate technology knowledge graphs, the method further includes: Obtain the preset modification cost coefficients for the target enterprise for different types of drawing editing operations. The modification cost coefficients include equipment purchase cost coefficients, process change cost coefficients, and downtime cost coefficients. For the top three candidate technology knowledge graphs, extract their corresponding node replacement operation list and node insertion operation list respectively. Multiply each operation in the node replacement operation list and the node insertion operation list by the corresponding transformation cost coefficient, and sum them up to obtain the estimated total technology transfer cost for each candidate technology knowledge graph. When multiple candidate technology knowledge graphs have the same number of minimum graph editing operations, the candidate technology knowledge graph with the smaller estimated total technology transfer cost is ranked higher.

[0010] Preferably, the technology transfer matching and service for the corresponding candidate technologies specifically includes: Extract all node insertion operations corresponding to the top-ranked candidate technology knowledge graph and generate a new equipment procurement list as hardware configuration service content. Extract all node replacement operations corresponding to the top-ranked candidate technology knowledge graph and generate a process flow change guide as part of the technical consulting service. Extract all edge insertion operations corresponding to the top-ranked candidate technology knowledge graph, generate a data interface docking solution as software integration service content; package the new equipment procurement list, the process flow change guide, and the data interface docking solution into a technology transfer implementation plan and output it to the target enterprise.

[0011] Preferably, after marking the corresponding technology node as a blocking node, the method further includes: Determine the attribute category of the blocked node, whereby the attribute category includes physically immutable attributes and logically mutable attributes; If the blocking node belongs to the physically immutable property, then the candidate technology knowledge graph containing the blocking node is marked as an infeasible solution, and the infeasible solution is removed from the step of sorting multiple candidate technology knowledge graphs. If the blocking node belongs to the logically variable attribute, then the candidate technology knowledge graph is retained and the step of determining the minimum number of graph edit operations on the enterprise technology knowledge graph required to eliminate all the blocking nodes and blocking edges continues.

[0012] Preferably, the node replacement operation in the minimum number of graph editing operations is specifically defined as: In the enterprise technology knowledge graph, an existing node with the same functional attributes as the blocked node but different parameter attributes is located; Modify the parameter attribute values ​​of the existing node to the parameter attribute values ​​of the blocked node; The node insertion operation is specifically defined as follows: when there is no node with the same functional attributes as the blocked node in the enterprise technology knowledge graph, a node that is completely identical to the blocked node is added at the corresponding logical position in the enterprise technology knowledge graph.

[0013] Preferably, the step of performing technology transfer matching and service for the corresponding candidate technologies further includes: A visual comparison chart is generated, in which the areas of the enterprise technology knowledge graph that require node replacement are highlighted in a first color; the areas of the enterprise technology knowledge graph that require node insertion are highlighted in a second color; and the areas of the enterprise technology knowledge graph that require simultaneous replacement and insertion are highlighted in a third color. The visual comparison chart is sent to the technology transfer service provider's terminal as part of the technology transfer risk assessment report.

[0014] The advantage of this invention over existing technologies lies in its ability to achieve a deep quantitative assessment of technology transfer adaptability by constructing an enterprise technology knowledge graph and a candidate technology knowledge graph, and utilizing graph matching and the computational logic of minimum graph edit operations. This invention goes beyond simply focusing on functional similarity; it delves into the obstacles encountered when new technologies are integrated into an enterprise's existing system—namely, incompatible nodes and edges. By calculating the minimum number of graph edit operations required to eliminate these obstacles, this invention can intuitively quantify the extent of changes an enterprise needs to make to adopt new technologies. This graph edit distance-based ranking mechanism ensures that the technologies recommended to enterprises not only meet functional requirements but also minimize implementation costs, thereby significantly reducing the risks and costs of technology transfer.

[0015] Furthermore, this invention addresses the issue of missing matches due to differences in wording in technical documents by combining natural language processing models for semantic vector matching, thereby improving recognition accuracy. By introducing the A* search algorithm, this invention can efficiently plan the optimal transformation path from the current state to the target state. Simultaneously, this invention improves sorting accuracy by combining cost coefficients for equipment procurement and process changes. Moreover, this invention can automatically generate specific implementation plans, including new equipment procurement lists, process change guidelines, and data interface integration schemes, based on graph editing operations. It also presents the transformation areas visually in the form of a comparative comparison chart, providing enterprise decision-makers with clear risk assessment criteria and implementation guidelines. Attached Figure Description

[0016] Figure 1 This is an overall flowchart of the method of the present invention; Figure 2 This is the construction diagram of the knowledge graph of this invention; Figure 3 This is a schematic diagram of the knowledge graph of this invention; Figure 4 This is a matching diagram of the knowledge graph of this invention; Figure 5 This is a schematic diagram illustrating the blocking node identification method of the present invention. Detailed Implementation

[0017] The specific embodiments of the present invention will now be described with reference to the accompanying drawings.

[0018] like Figure 1 As shown, this invention constructs a digital map to structurally compare the abstract current state of an enterprise with candidate technologies, quantifies the gaps and costs in technology implementation, and thus generates an executable implementation plan.

[0019] More specifically, such as Figure 2 and Figure 3As shown, this embodiment first establishes a standardized knowledge graph construction paradigm, aiming to transform unstructured technical documents into computer-understandable and computable graph-structured data. This step forms the basis for subsequent difference analysis and path planning.

[0020] In its construction process, this invention breaks down the entire lifecycle of technology implementation into atomic elements. Specifically, entity nodes are strictly divided into three categories: The first category is process nodes, which represent specific action units in the production or business process. For example, in the welding process of automotive parts, laser alignment, arc ignition, and molten pool cooling are all independent process nodes.

[0021] The second category is hardware device nodes, which refer to the physical devices or tools required to perform the above-mentioned processes. Examples include six-axis robotic arms, laser generators, and wire feeders.

[0022] The third category is process parameter nodes, which are quantitative constraints on processes or equipment. Examples include a current intensity of 200 amperes, a shielding gas flow rate of 15 liters per minute, and a welding speed of 5 millimeters per second.

[0023] To accurately describe the relationships between nodes, this invention defines three types of directed edges: A temporal edge is used to connect two process nodes, representing the sequential order of business processes. For example, an edge pointing from laser alignment to the arc ignition represents irreversible temporal logic.

[0024] Edges are used to describe the dependency relationship between hardware devices and processes. They are usually formed by process nodes pointing to hardware device nodes, indicating that the process depends on the operation of that specific device.

[0025] Attribute edges are used to connect process parameters and process nodes, indicating that the parameter is an operating condition that the process must meet.

[0026] For target companies, the system accesses their enterprise resource planning (ERP) or manufacturing execution (MEP) systems for technical documents, work instructions, and equipment ledgers. For candidate new technologies, the system retrieves their patent specifications, technical white papers, or implementation manuals.

[0027] To transform these texts into graphs, this invention employs a deep learning model based on the BERT architecture for information extraction. The specific model architecture includes a 12-layer Transformer encoder.

[0028] First, named entity recognition is performed. The text is input into a pre-trained BERT model to obtain word-level vector representations. These are then fed into a Bidirectional Long Short-Term Memory (BiLSTM) network to capture sequence features, and finally, a Conditional Random Field (CRF) layer is used to output a sequence of entity labels. For example, the model might identify 3000 watts as a process parameter in a 3000-watt laser, and the laser as a hardware device.

[0029] Then, relation extraction is performed. For the identified entity pairs, they and the context sentences containing them are input into the model, and the classification layer determines whether there are temporal, compositional, or attribute relationships between them.

[0030] Through the above steps, the unstructured text is transformed into two structured graphs: an enterprise technology knowledge graph representing the current state of the enterprise. and a knowledge graph of candidate technologies representing new technological solutions .

[0031] like Figure 4 and Figure 5 As shown, after the system is built, it needs to identify the gap between the company's current situation and new technologies, that is, to find the bottlenecks. This is essentially a process of graph matching and difference analysis.

[0032] To overcome the matching difficulties caused by synonyms or differences in expression, this invention does not employ simple keyword matching, but instead performs calculations based on semantic vector space. The system utilizes a Sentence-BERT model fine-tuned with industrial corpora to map each node in the two graphs into a 768-dimensional high-dimensional dense vector.

[0033] Suppose a node in the candidate technology map The vector is A node in the enterprise technology map The vector is The system calculates the cosine similarity between the two. The formula is as follows: ; in represents the dot product of vectors, and |||| represents the magnitude of the vector.

[0034] The system sets a similarity threshold. In this embodiment, the value is taken as 0.85. For each node in the candidate graph, the system traverses all nodes in the enterprise graph to calculate the similarity. If for a certain candidate node... There are no enterprise nodes. Make This means that the company's existing system completely lacks this technological element. In this case, the system will... Marked as a blocked node.

[0035] For example, if a new technology requires the use of fiber lasers, but the company currently uses carbon dioxide lasers, and the semantic model determines that the two have low similarity and cannot be substituted for each other, then the fiber laser is marked as a blocking node.

[0036] After identifying the blocked nodes, in order to avoid recommending technologies that are impossible to implement to enterprises (such as equipment modifications that require the removal of load-bearing walls), the system introduced an attribute category judgment mechanism.

[0037] The system has a built-in attribute classifier that divides blocked nodes into physically immutable attributes and logically mutable attributes.

[0038] Physically immutable properties refer to elements that cannot be changed due to limitations imposed by the site's physical structure, mandatory legal and regulatory requirements, or extremely high modification costs. Examples include factory floor height restrictions, maximum electrical load limits, or specific wastewater discharge permit restrictions.

[0039] Logically variable attributes refer to elements that can be changed by purchasing equipment, modifying software parameters, or training employees. Examples include welding voltage, production line cycle time, or control algorithm version.

[0040] If the candidate technology graph contains blocking nodes with physically immutable attributes, the system directly determines that the technology solution is infeasible and removes it from the subsequent ranking list, thereby saving computational resources and improving the accuracy of the recommended solution. Conversely, if the blocking nodes are all logically mutable attributes, the solution is retained and proceeds to the next stage of computation.

[0041] Node matching only addresses the question of what exists, while edge matching addresses the question of how to do it. After confirming the node matching relationships, the system further verifies the consistency of the edges.

[0042] Assume there is an edge in the candidate graph. , connected nodes and nodes And nodes and The corresponding matching nodes have been found in the enterprise graph. and At this point, the system checks... and Does an edge exist between them? .

[0043] The judgment criteria include two dimensions: Attribute type consistency, if If it's a temporal edge, then... It must also be a sequential edge. If Require Prior to Execution, and the current state of enterprises and If there is no temporal relationship between them, or the relationship type is composition relationship, then a mismatch is determined.

[0044] Consistency in direction, if From point to ,and From point to This indicates a conflict in the process logic.

[0045] Any edge that meets the above mismatch criteria is marked as a blocking edge. This means that the enterprise's existing logical processes cannot support the operation of new technologies and require process reengineering.

[0046] In order to objectively evaluate which new technology is most suitable for the enterprise, this embodiment introduces the concept of graph edit distance and models the transformation process as a problem of solving the minimum number of graph edit operations.

[0047] This invention views technology transfer as the process of creating a corporate technology map. Gradually transform into a map that can accommodate candidate technologies The process involves the system constructing a state-space search tree. The root node represents the initial state of the enterprise graph. The target state represents a state isomorphic to the candidate technology map. .

[0048] The rule for generating child nodes is that for each blocked node or blocked edge in the current state graph, a graph editing operation (node ​​replacement, node insertion, edge replacement, edge insertion) is applied to generate a new graph state as a child node.

[0049] To give the algorithm practical physical meaning, this invention provides rigorous engineering definitions for the four operations: Node replacement corresponds to upgrading or adjusting existing resources. The algorithm locates a node in the enterprise graph that has the same functional attributes as the blocked node (e.g., both are heating devices), but different parameter attributes (e.g., one is electric heating, and the other is infrared heating), and modifies its parameter values ​​to the target values.

[0050] Node insertion corresponds to the introduction of entirely new resources. When a certain type of function is completely absent from the enterprise graph, the algorithm adds a new node at a logical location. For example, adding a vision inspection camera to an existing production line.

[0051] Simultaneous replacement and insertion correspond to changes in the process flow or the establishment of data interfaces. For example, changing the order of two processes, or establishing a new data transmission link between equipment and the central control system.

[0052] Because the combination of operations in graph editing explodes exponentially, the system employs the A* search algorithm to quickly find the minimum number of operations. The core of the algorithm lies in the evaluation function. .

[0053] Represents the transition from the initial state to the current state. The number of operations already performed. In this stage, the path weight of all four types of operations is set to 1, meaning that each operation is considered to have an equal cost at the abstract level.

[0054] It is a heuristic function used to predict the state from the current state. The minimum cost required to reach the target state. This embodiment uses the total number of remaining unmatched blocked nodes and blocked edges in the current graph as the minimum cost. The value of . This heuristic function is admissible and guarantees that the optimal solution will be found.

[0055] The algorithm continuously retrieves from the priority queue The minimum state is expanded until the target state is reached. At this point, the total number of operations accumulated along the path is the minimum number of graph edit operations.

[0056] While the minimum number of graph editing operations reflects the magnitude of the technical differences, it does not take into account the actual cost. Therefore, after the initial sorting, the system can further introduce weights for a secondary, refined sorting.

[0057] The system allows businesses to preset transformation cost coefficients based on their own circumstances. These coefficients transform abstract graph operations into monetary or time costs.

[0058] Equipment procurement cost coefficient : Corresponds to node insertion operation. This coefficient is usually set to a large value, such as between 1,000 and 100,000, depending on the average price of equipment in the industry.

[0059] Process change cost coefficient : Corresponds to node replacement operation. This coefficient reflects the cost of adjusting parameters or making minor modifications, and its value can range from 100 to 5000.

[0060] Downtime cost coefficient : Corresponding edge operations. Process reorganization (adding or deleting edges) usually means production line shutdown for debugging. The value range can be between 500 and 20000.

[0061] For the top three candidate techniques ranked by the minimum number of operations, the system extracts the operation list along their optimal path. Let the number of node insertions be... The number of node replacements is The total number of operations (including insertion and replacement) is .

[0062] Estimated total cost of technology transfer The calculation formula is as follows: ; The system according to The top three options will be reordered. If two options have the same number of operations, but in actual implementation, option A only requires parameter modification (low cost), while option B requires purchasing new equipment (high cost), then option A will be moved to a higher position.

[0063] The final output of this invention is not merely a recommendation, but a complete package of technology transfer implementation schemes.

[0064] The system analyzes the optimal editing path corresponding to the top-ranked candidate technology, mapping the abstract operation to a specific service document: The hardware configuration service includes retrieving all node insertion operations. For example, if the operation is to insert an [infrared sensor] node, the system automatically queries the supplier database and generates a "New Equipment Procurement List" that includes the equipment name, specifications, and recommended suppliers.

[0065] Technical consulting services include extracting all node replacement operations. For example, if the operation is to replace the [heating temperature] from 150 degrees to 180 degrees, the system will generate a "Process Flow Change Guide," which clearly indicates the adjustment targets and debugging methods for the original process parameters.

[0066] The software integration service includes extracting all edge insertion operations. For example, if the operation is to insert a data edge from the [PLC controller] to the [cloud server], the system will generate a "Data Interface Connection Scheme" that defines the data transmission protocol, baud rate, and interface format.

[0067] To help corporate decision-makers intuitively understand the difficulties and risks of technology transfer, the system generates a visual comparison chart.

[0068] This diagram uses the company's original technology flowchart as a base map and uses color coding to represent the areas to be modified: The red highlighted area (first color) indicates the region where node replacement needs to be performed. This means that this area is a high-risk zone, and the quality fluctuations caused by parameter adjustments need to be closely monitored.

[0069] The yellow highlighted area (second color) indicates the region where node insertion needs to be performed. This visually demonstrates the installation location of the new equipment, assisting in factory layout planning.

[0070] The blue highlighted area (third color) indicates the region where simultaneous replacement and insertion are required. This represents a change in the path of information and logistics.

[0071] This visualization chart, together with the three service documents mentioned above, constitutes the "Technology Transfer Risk Assessment Report," which is sent to the technology transfer service provider's terminal via the network as a basis for expert intervention and guidance decisions.

[0072] Through the above implementation methods, the present invention achieves intelligent processing of the entire process from demand identification, technology matching, cost estimation to solution generation, and solves the core pain points of difficult supply and demand matching, opaque transformation costs, and unclear implementation paths in the traditional technology transfer process.

[0073] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for demand identification and service configuration of technology transfer services, characterized in that, Includes the following steps: Based on a pre-defined knowledge graph construction method, the overall technology implementation process of the target enterprise is constructed into an enterprise technology knowledge graph. Following the same knowledge graph construction method as the enterprise technology knowledge graph, multiple candidate new technologies are constructed into candidate technology knowledge graphs. Specifically, the pre-defined knowledge graph construction method involves defining each process step, each hardware device, and each process parameter in the technology implementation process as an entity node; defining the logical flow relationship between processes as a temporal edge; defining the subordinate relationship between processes and hardware devices as a component edge; and defining the constraint relationship of process parameters on processes as attribute edges. By extracting entity nouns and verbs from the target enterprise's technical documents and mapping them to entity nodes, temporal edges, component edges, and attribute edges, the enterprise technology knowledge graph is generated. Similarly, by extracting entity nouns and verbs from the technical specifications of the candidate new technologies and mapping them to entity nodes, temporal edges, component edges, and attribute edges, the candidate technology knowledge graph is generated. Graph matching is performed between the candidate technology knowledge graph and the enterprise technology knowledge graph. When a node in the candidate technology knowledge graph cannot be matched with the enterprise technology knowledge graph, the corresponding technology node is marked as a blocked node. When a conditional edge in the candidate technology knowledge graph cannot be matched with the enterprise technology knowledge graph, the corresponding conditional edge is marked as a blocked edge. The minimum number of graph editing operations required to eliminate all the blocked nodes and edges in the enterprise's technical knowledge graph is determined. The graph editing operations include node replacement, node insertion, edge replacement, and edge insertion. When determining the minimum number of graph editing operations, the total number of remaining unmatched blocked nodes and edges in the current state is used as the heuristic function value for optimal path search. The heuristic function has acceptability to ensure that the determined minimum number of graph editing operations is the globally optimal solution. Multiple candidate technology knowledge graphs are sorted, with the candidate technology knowledge graph having fewer minimum graph editing operations being sorted higher, and the corresponding candidate technology technology transfer matching and services are performed based on the sorting. The technology transfer matching and service for the corresponding candidate technologies includes: extracting all node insertion operations corresponding to the top-ranked candidate technology knowledge graph to generate a new equipment procurement list as hardware configuration service content; extracting all node replacement operations corresponding to the top-ranked candidate technology knowledge graph to generate a process flow change guide as technical consulting service content; extracting all edge insertion operations corresponding to the top-ranked candidate technology knowledge graph to generate a data interface docking solution as software integration service content; and packaging the new equipment procurement list, the process flow change guide, and the data interface docking solution into a technology transfer implementation plan and outputting it to the target enterprise.

2. The method for demand identification and service configuration of technology transfer services according to claim 1, characterized in that, The specific steps for graph matching between candidate technology knowledge graphs and enterprise technology knowledge graphs include: Each node in the candidate technology knowledge graph is transformed into a candidate node semantic vector using a pre-trained natural language processing model, and each node in the enterprise technology knowledge graph is transformed into an enterprise node semantic vector. Calculate the cosine similarity between the semantic vector of each candidate node and the semantic vector of each enterprise node; When the cosine similarity between a node in the candidate technology knowledge graph and any node in the enterprise technology knowledge graph is less than a preset similarity threshold, the node is determined to be unmatchable and is marked as the blocking node.

3. The method for demand identification and service configuration of technology transfer services according to claim 2, characterized in that, The specific steps for marking blocked edges include: In the candidate technology knowledge graph, the connection edge between two matched nodes is identified as the edge to be verified. Search for two enterprise nodes in the enterprise technology knowledge graph that correspond to the two matched nodes; Determine whether there is a connection edge between the two enterprise nodes that matches the attribute type of the edge to be verified. If there is no connecting edge between the two enterprise nodes that matches the attribute type of the edge to be verified, or if the direction of the connecting edge between the two enterprise nodes is opposite to the direction of the edge to be verified, then the edge to be verified is marked as the blocking edge.

4. The method for demand identification and service configuration of technology transfer services according to claim 1, characterized in that, The steps for determining the minimum number of image editing operations specifically include: Construct a state space search tree, taking the initial state of the enterprise technology knowledge graph as the root node and the graph state after eliminating one of the blocking nodes or one of the blocking edges as the child nodes. Define the path weights of the four operations—node replacement, node insertion, edge replacement, and edge insertion—as 1. Using A The search algorithm searches the state space search tree for the shortest path from the initial state to the target state that is isomorphic to the candidate technology knowledge graph. The total number of node replacements, node insertions, edge replacements, and edge insertions along the shortest path is counted, and this total number is determined as the minimum number of graph editing operations.

5. The method for demand identification and service configuration of technology transfer services according to claim 4, characterized in that, Following the step of ranking multiple candidate technology knowledge graphs, the following also includes: Obtain the preset modification cost coefficients for the target enterprise for different types of drawing editing operations. The modification cost coefficients include equipment purchase cost coefficients, process change cost coefficients, and downtime cost coefficients. For the top three candidate technology knowledge graphs, extract their corresponding node replacement operation list and node insertion operation list respectively. Multiply each operation in the node replacement operation list and the node insertion operation list by the corresponding transformation cost coefficient, and sum them up to obtain the estimated total technology transfer cost for each candidate technology knowledge graph. When multiple candidate technology knowledge graphs have the same number of minimum graph editing operations, the candidate technology knowledge graph with the smaller estimated total technology transfer cost is ranked higher.

6. The method for demand identification and service configuration of technology transfer services according to claim 1, characterized in that, After marking the corresponding technology node as a blocked node, the method further includes: Determine the attribute category of the blocked node, whereby the attribute category includes physically immutable attributes and logically mutable attributes; If the blocking node belongs to the physically immutable property, then the candidate technology knowledge graph containing the blocking node is marked as an infeasible solution, and the infeasible solution is removed from the step of sorting multiple candidate technology knowledge graphs. If the blocking node belongs to the logically variable attribute, then the candidate technology knowledge graph is retained and the step of determining the minimum number of graph edit operations on the enterprise technology knowledge graph required to eliminate all the blocking nodes and blocking edges continues.

7. The method for demand identification and service configuration of technology transfer services according to claim 1, characterized in that, The node replacement operation in the minimum number of graph editing operations is specifically defined as follows: In the enterprise technology knowledge graph, an existing node with the same functional attributes as the blocked node but different parameter attributes is located; Modify the parameter attribute values ​​of the existing node to the parameter attribute values ​​of the blocked node; The node insertion operation is specifically defined as follows: when there is no node with the same functional attributes as the blocked node in the enterprise technology knowledge graph, a node that is completely identical to the blocked node is added at the corresponding logical position in the enterprise technology knowledge graph.

8. The method for demand identification and service configuration of technology transfer services according to claim 1, characterized in that, The steps of performing technology transfer matching and services for corresponding candidate technologies also include: A visual comparison chart is generated, in which the areas of the enterprise technology knowledge graph that require node replacement are highlighted in a first color; the areas of the enterprise technology knowledge graph that require node insertion are highlighted in a second color; and the areas of the enterprise technology knowledge graph that require simultaneous replacement and insertion are highlighted in a third color. The visual comparison chart is sent to the technology transfer service provider's terminal as part of the technology transfer risk assessment report.

Images