Business processing method and system

Abstract

The application discloses a business processing method and system. The method comprises the following steps: based on the current business processing node of a to-be-processed business, tracing back to the historical business processing node in a directed graph set to determine the business execution path of the current business processing node; the directed graph set comprises the business processing nodes of multiple businesses and the execution sequence between the business processing nodes; executing the business processing logic of the current business processing node to obtain the business processing result of the current business processing node; inputting the business execution path and the business processing result into a business processing model for business analysis to obtain the next business processing node of the to-be-processed business, and jumping to the next business processing node to execute the corresponding business processing logic until the next business processing node is the terminal node in the directed graph set. The method can avoid missing business processing nodes and ensure the integrity and orderliness of business processing.

Description

Technical Field

[0001] This application relates to the field of large language models, and in particular to a business processing method and system. Background Technology

[0002] With the development of artificial intelligence technology, large language models have also evolved. Due to their powerful natural language processing capabilities, large language models are widely used in various fields such as intelligent customer service, content creation, healthcare, and cybersecurity. Especially for business scenarios requiring strict adherence to standard operating procedures (SOPs), large language models can quickly and effectively resolve users' business needs while saving human resources.

[0003] Currently, when processing business based on large language models, the business process described in natural language is often directly input into the large language model. The large language model then uses the business process as a reference and advances the process based on preset paths and predefined rules. This approach suffers from issues such as missing business processing nodes and limitations in the business processing flow, which affects the accuracy of the results and makes it difficult for the results to meet user needs. Summary of the Invention

[0004] In view of the above-mentioned defects or deficiencies in the prior art, it is desirable to provide a business processing method and system that can avoid missing business processing nodes and ensure the integrity and orderliness of business processing.

[0005] Firstly, this application provides a business processing method. The method includes: Based on the current business processing node of the pending business, the historical business processing nodes are traced back in the directed graph set to determine the business execution path of the current business processing node; the directed graph set includes business processing nodes of various businesses and the execution order between each business processing node; Execute the business processing logic of the current business processing node and obtain the business processing result of the current business processing node; The business execution path and business processing result are input into the business processing model for business parsing, to obtain the next business processing node of the business to be processed, and jump to the next business processing node to execute the corresponding business processing logic, until the next business processing node is the end node in the directed graph set.

[0006] In conjunction with the first aspect, in one possible implementation, the construction process of the directed graph set includes: obtaining natural language description texts of business processing flows for various services; performing grammatical transformation and node annotation on the natural language description texts based on a preset graph language to obtain graph language description texts of various business processing flows; node annotations include processing logic annotations, attribute annotations, and connection relationship annotations for each business processing node; parsing the graph language description texts to generate a directed graph set corresponding to various business processing flows; the directed graph set includes the upstream node set, downstream node set, and annotation information for each business processing node.

[0007] In conjunction with the first aspect, in one possible implementation, determining the business execution path of the current business processing node includes: determining the downstream node of the current business processing node based on the execution order of the current business processing node; and constructing the business execution path based on the current business processing node, the historical business processing nodes executed before the current business processing node, and the downstream node of the current business processing node.

[0008] In conjunction with the first aspect, in one possible implementation, determining the business execution path of the current business processing node includes: for each historical business processing node, determining the adjacent nodes of the historical business processing node; the adjacent nodes are the same as the parent nodes of the historical business processing nodes; and constructing the business execution path based on the current business processing node, the historical business processing nodes executed before the current business processing node, the downstream nodes, and the adjacent nodes.

[0009] In conjunction with the first aspect, in one possible implementation, the business processing result includes the user semantics expressed by the user during the business processing process of the current business processing node. The business parsing of the business processing model includes: matching the user semantics with the business processing logic of the current business processing node, the business processing logic of the historical business processing node, and the business processing logic of the downstream node; if the user semantics do not match the business processing logic of the current business processing node, the business processing logic of the historical business processing node, and the business processing logic of the downstream node, the user semantics are matched with the business processing logic of the adjacent node, and the matched adjacent node is determined as the next business processing node of the current business processing node.

[0010] In conjunction with the first aspect, in one possible implementation, determining the business execution path of the current business processing node includes: for each historical business processing node, determining the adjacent nodes and bypass nodes of the historical business processing node; the adjacent nodes are the same as the parent nodes of the historical business processing nodes, and the bypass nodes are the business processing nodes in the branches where the adjacent nodes are located; and constructing the business execution path based on the current business processing node, the historical business processing nodes executed before the current business processing node, the downstream nodes, the adjacent nodes, and the bypass nodes.

[0011] In conjunction with the first aspect, in one possible implementation, the business parsing result output by the business processing model also includes a response text. Before jumping to the next business processing node, the method further includes: matching the response text with risk keywords in a preset risk keyword set; if there is a risk keyword that matches the response text, then determining the target response text based on the matched risk keyword and risk processing rule information; the risk processing rule information includes the correspondence between different risk keywords and standard response texts; and replacing the response text based on the target response text.

[0012] In conjunction with the first aspect, in one possible implementation, the business processing method further includes: after completing the business processing logic of the current business processing node, updating the execution status of the current business processing node, and sending the execution status to the terminal device to instruct the terminal device to update the visualization style information of the current business node in the visualization directed graph based on the execution status; the visualization directed graph is a visualization flowchart of a set of directed graphs corresponding to various businesses.

[0013] In conjunction with the first aspect, in one possible implementation, the business processing result includes the collected business data, and the method further includes: verifying the business data of the current business processing node to obtain the verification result; and inputting the business execution path, business processing result, and verification result into the business processing model for business parsing.

[0014] Secondly, this application also provides a business processing system. The system includes a terminal device and a server, the server being used to implement the business processing methods described in the first aspect and any one of the claims in the first aspect.

[0015] Thirdly, this application also provides a computer program product. This computer program product includes a computer program that, when executed by a processor, implements the method described in the first aspect.

[0016] This application provides a business processing method and system. A directed graph set is pre-constructed based on business processing nodes and their execution order for various businesses. When executing a business to be processed, the execution path of the current business processing node is determined by tracing back historical business processing nodes in the directed graph set, and the business processing logic of the current business processing node is executed to obtain the business processing result. The execution path and the business processing result are input into a business processing model for business parsing to obtain the next business processing node for the business to be processed. The system then jumps to the next business processing node to execute the corresponding business processing logic until the next business processing node is the end node in the directed graph set. The method provided in this application can refer to the pre-constructed directed graph set of business processing nodes and their execution order to execute each business processing node involved in the business to be processed, avoiding omissions of business processing nodes and ensuring the completeness and orderliness of business processing. Furthermore, this application embodiment uses the business execution path and business processing result of the current business processing node as model input for business analysis, avoiding inputting the entire business processing flow into the large language model for analysis. While ensuring the richness of the data input to the model, it effectively reduces the amount of data, avoids affecting the attention of the large language model, and improves the efficiency and accuracy of business parsing based on the large language model. Attached Figure Description

[0017] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a schematic diagram of the architecture of a business processing system in one embodiment; Figure 2 This is a flowchart illustrating a business processing method in one embodiment; Figure 3 A visual flowchart of a directed graph set of multiple services in one embodiment; Figure 4 This is another flowchart illustrating the business processing method in one embodiment; Figure 5 This is another flowchart illustrating the business processing method in one embodiment; Figure 6 This is another flowchart illustrating the business processing method in one embodiment; Figure 7 This is another flowchart illustrating the business processing method in one embodiment; Figure 8 This is a schematic diagram of the business execution path of the current business processing node in one embodiment; Figure 9 This is another flowchart illustrating the business processing method in one embodiment; Figure 10 This is another flowchart illustrating the business processing method in one embodiment; Figure 11 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0018] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0019] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The present application will now be described in detail with reference to the accompanying drawings and embodiments. Furthermore, the term "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. The terms "first" and "second," etc., in the specification and claims of the embodiments of this application are used to distinguish different objects, not to describe a specific order of objects.

[0020] With the development of artificial intelligence technology, large language models have also evolved. Due to their powerful natural language processing capabilities, large language models are widely used in various fields such as intelligent customer service, content creation, healthcare, and cybersecurity. Especially for business scenarios requiring strict adherence to standard operating procedures (SOPs), large language models can quickly and effectively resolve users' business needs while saving human resources.

[0021] Currently, when processing business logic based on large language models, the business process described in natural language is often directly input into the large language model. The large language model then uses the business process as a reference and proceeds according to preset paths and predefined rules. In this approach, the large language model often treats the business process as reference information, leading to the omission of business processing nodes (e.g., validation nodes), which affects the reliability and accuracy of the processing results. Furthermore, the business processing method based on preset paths suffers from limitations in the business process flow, making it difficult to adapt to different user needs, which also affects the accuracy of the processing results.

[0022] Based on this, embodiments of this application provide a business processing method and system that can avoid missing business processing nodes and ensure the integrity and orderliness of business processing.

[0023] The business processing method provided in this application embodiment can be applied to a server, or to other applications such as... Figure 1 In the business processing system shown, which consists of terminal device 10 and server 20, server 20 can execute the processing flow of pending business and interact with the user through terminal device 10 during the business processing. Terminal device 10 communicates with server 20 via a network. Terminal device 10 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, etc., and server 20 can be implemented as a standalone server or a server cluster consisting of multiple servers.

[0024] In one embodiment, such as Figure 2 As shown, a business processing method is provided, which is applied to... Figure 1 Taking server 20 as an example, the explanation includes the following steps: Step 101: Based on the current business processing node of the business to be processed, backtrack the historical business processing nodes in the directed graph set to determine the business execution path of the current business processing node.

[0025] The directed graph set includes business processing nodes for various services and the execution order between these nodes. The services used to construct the directed graph set can be those that require strict adherence to standard operating procedures (SOPs), such as installation scheduling and maintenance scheduling.

[0026] In this embodiment, to achieve orderly business processing, the business processing nodes involved in the business processing flow of various businesses can be summarized first. Then, according to the execution order of each business processing node, a directed graph set corresponding to each business can be constructed, and this directed graph set can be stored locally on server 20. The directed graph set includes the annotation information of each business processing node, such as node identifier, node type (e.g., start node, end node, condition judgment node, etc.), and business processing logic (blocking the business processing flow, inquiring about product type, sending SMS, satisfaction evaluation, etc.). For example, the business processing nodes involved in product installation and product repair businesses (e.g., inquiring about service type, inquiring about product category, inquiring about willingness to self-repair, judging whether the stove opening or countertop condition is met, etc.) can be constructed according to the execution order as follows: Figure 3 The set of directed graphs shown is stored locally on server 20.

[0027] Server 20 receives service requests initiated by users. For example, a user clicks "Schedule Installation" through the service platform provided by terminal device 10; a user sends a message "My range hood is not working and needs repair" through the interactive interface provided by terminal device 10; a user calls the product robot customer service through terminal device 10 to request a service request to install the range hood. Terminal device 10 responds to the user's service request by sending the service request to server 20.

[0028] After receiving a business request, server 20 responds by starting from the starting node in the directed graph set (e.g., the business processing node "determine service type" in the diagram), using the starting node as the current business processing node, and executing the business processing logic of the current business processing node. After completing the processing of the current business processing node, it jumps to the next business processing node, updates the next business processing node to the current business processing node, and executes the corresponding business processing logic, until the next business processing node is the end node in the directed graph set, i.e., the terminal node, such as "send self-service guidance SMS", "end", "suggest making an appointment after meeting the conditions" in the diagram. Then, the processing flow of the pending business indicated by the business request is determined to be completed.

[0029] Specifically, when the business processing flow of a pending business jumps to the current business processing node, the server can start from the current business processing node and trace back through the parent nodes to the historical business processing nodes that have been executed before the current business processing node. Until traversing back to the starting node, the server 20 can determine the business execution path corresponding to the current business processing node based on the current business processing node, all historical business processing nodes corresponding to the current business processing node, and the execution order among these business processing nodes, and extract the business execution path from the directed graph set. For example, if the pending business execution reaches the current business processing node "Can it be self-resolved?", the server 20 determines that the parent node of "Can it be self-resolved?" is "Inquire about the fault phenomenon", the parent node of "Inquire about the fault phenomenon" is "Inquire about the product model", and so on, until it is determined that the historical business processing nodes for "Can it be self-resolved?" include "Inquire about the fault phenomenon", "Inquire about the product model", "Inquire about the product category", and "Determine the service type". Based on the execution order of the above nodes, the business execution path corresponding to the current business processing node "Can it be self-resolved?" can be determined as: Determine the service type – Inquire about the product category – Inquire about the product model – Inquire about the fault phenomenon – Can it be self-resolved?

[0030] Step 102: Execute the business processing logic of the current business processing node and obtain the business processing result of the current business processing node.

[0031] In this embodiment of the application, when the business processing flow of the business to be processed jumps to the current business processing node, the business processing logic of the current business processing node is obtained from the directed graph set, and the business processing logic is executed to obtain the business processing result.

[0032] In one possible implementation, if the business processing type indicated by the current business processing node is an inquiry, the business processing logic can initiate a session (via voice call or interactive interface) based on the preset slot information (e.g., product model) of the current business processing node. Then, the user's dialogue text is recorded, and slot extraction is performed on the user's dialogue text for semantic understanding. The resulting business processing result can include at least one of user semantics and business data. For example, if the current business processing node is "inquiring about product model," the server 20 can output the text "What is the model of the range hood you want to repair?" through the interactive interface of the terminal device 10. Then, the user's dialogue text "Model is XXX" is obtained, and semantic understanding and slot extraction are performed on the user's dialogue text to obtain the business data, namely "Product model to be repaired: XXX". Here, the slot information represents the business data that the current business processing node needs to obtain.

[0033] In one possible implementation, if the business processing type indicated by the current business processing node is not an inquiry type, the corresponding business processing logic can be executed directly. For example, the business processing logic for "generating a work order" could be based on historically acquired business data, generating an appointment work order according to a preset template, and sending it back to the user terminal 20; the business processing logic for "sending a self-service guidance SMS" could be based on historically acquired product models and fault types, querying the corresponding troubleshooting guidance text, and sending it to the user terminal 20 via SMS.

[0034] Step 103: Input the business execution path and business processing result into the business processing model for business parsing, obtain the next business processing node of the business to be processed, and jump to the next business processing node to execute the corresponding business processing logic until the next business processing node is the end node in the directed graph set.

[0035] Understandably, the business execution path includes the business processing results of each historical business processing node corresponding to the current business processing node. Therefore, the business execution path can represent the context information of the current business processing node, avoiding the repeated execution of already processed business processing nodes or the repeated querying of already acquired business data in subsequent business processing flows, while ensuring that the business parsing of the business processing model is based on real and rich data.

[0036] In this embodiment, the business processing flow of the business to be processed jumps to the current business processing node. After obtaining the business execution path and business processing result of the current business processing node, the business execution path and business processing result can be input into a pre-trained business processing model. The business processing model is a Large Language Model (LLM), which can be specifically trained using a large language model from the business processing domain (such as GPT, AskBot, etc.) as the initial model. Using a large model from the business processing domain as the initial model can effectively improve the efficiency and accuracy of business processing.

[0037] The business processing model performs business parsing based on the business execution path and business processing result, and outputs the business parsing result of the current business processing node. This result may include the next business processing node to be jumped to. It should be noted that the next business processing node is the next business processing node used to further advance the business process to be processed, and not a downstream node connected to the current business node. The next business processing node may be a downstream node of the current business processing node, a historical business processing node of the current business processing node, still the current business processing node, or a bypass node in the directed graph set other than the business execution path corresponding to the current business processing node.

[0038] For example, if server 20 determines that the product fault reported by the user can be resolved by the user themselves at the current business processing node "Can it be resolved by self-service?", it will jump to the downstream node "Will you be willing to resolve it yourself?" to ask the user if they are willing to resolve the product fault themselves according to the guidance information. For example, if during the conversation at the current business processing node "Inquire about the fault phenomenon", the server receives the user's dialogue "I just said the wrong product model, it should be XXX", it will jump to the historical business processing node "Inquire about the product model" to obtain the product model to be repaired again. For example, if during the conversation at the current business processing node "Inquire about the fault phenomenon", the server receives the user's dialogue "I forgot to mention the fault phenomenon, there is also XXX", it will remain at the current business processing node and again obtain the business execution path, business processing result, and input business processing model for the current business processing node. For example, if during the conversation at the current business processing node "Inquire about the product model", after initiating the conversation "What is the model of the range hood you want to repair?", and receives the user's dialogue "I said the wrong one before, I want to schedule the installation of a newly purchased range hood, not the repair of a range hood", it will jump to the bypass node "Confirm the installation category" to continue the business processing flow.

[0039] In this embodiment, server 20 jumps to the next business processing node output by the business processing model, updates the next business processing node to the current business processing node, and continues to obtain the business execution path, execute the business processing logic to obtain the business processing result, and input the business processing model. Until the next business processing node (i.e., the updated current business processing node) is the final node, the business processing flow of the pending business is completed after the business processing logic of the final node is executed.

[0040] In one possible implementation, in addition to the next business processing node, the business parsing result output by the business processing model may also include a response text. For example, during the session of the current business processing node, if the user dialogue "How much is the repair cost?" is obtained, the business processing model can output the response text "The specific cost will be determined by the repair technician based on the charging standard after on-site inspection"; for example, during the session of the current business processing node, if the user dialogue "Can this fault be repaired?" is obtained, the business processing model can output the response text "The repair result is subject to the on-site judgment of the repair technician".

[0041] The method provided in this application can pre-construct a directed graph set based on the business processing nodes of various services and the execution order between nodes. When executing a service to be processed, the method determines the business execution path of the current business processing node by tracing back historical business processing nodes in the directed graph set based on the current business processing node, and executes the business processing logic of the current business processing node to obtain the business processing result of the current business processing node. The business execution path and business processing result are input into the business processing model for business parsing to obtain the next business processing node of the service to be processed, and jumps to the next business processing node to execute the corresponding business processing logic, until the next business processing node is the end node in the directed graph set. The method provided in this application can refer to the pre-constructed directed graph set of business processing nodes and execution order to execute each business processing node involved in the service to be processed, avoiding omission of business processing nodes and ensuring the integrity and orderliness of business processing. Furthermore, this application embodiment uses the business execution path and business processing result of the current business processing node as model input for business analysis, avoiding inputting the entire business processing flow into the large language model for analysis. While ensuring the richness of the data input to the model, it effectively reduces the amount of data, avoids affecting the attention of the large language model, and improves the efficiency and accuracy of business parsing based on the large language model.

[0042] In one embodiment, a process for constructing a set of directed graphs is provided, which may include, for example: Figure 4 The steps shown are as follows: Step 201: Obtain natural language description text of business processing flows for various business operations, and perform syntax conversion and node annotation on the natural language description text based on a preset chart language to obtain chart language description text of various business processing flows.

[0043] Among them, node annotations include processing logic annotations, attribute annotations, and connection relationship annotations for each business processing node.

[0044] In this embodiment, natural language descriptions of various business processes pre-stored locally on server 20 can be retrieved. Then, the natural language descriptions are parsed and converted using a preset graph language, transforming them into graph language descriptions. For example, the natural language descriptions can be written as Mermaid text using Mermaid syntax.

[0045] During the syntax conversion process, the natural language description text also needs to be parsed to analyze the business processing nodes it indicates and the execution order between these nodes. Node annotation is then performed using graph language syntax tools, ultimately resulting in graph language description text with node annotations for various business processes. For example, during syntax conversion, tools such as `classDef` in Mermaid syntax are used to annotate the attributes (node ​​identifier, node type, etc.), business processing logic (which may include slot information), and connection relationships between nodes (i.e., the execution order of business processing nodes) of each business processing node, resulting in annotated Mermaid text.

[0046] In one possible implementation, the default graph language could be Graphviz (DOT language), JSON syntax language, YAML syntax language, domain-specific language (DSL), etc.

[0047] Step 202: Parse the chart language description text and generate a set of directed graphs corresponding to various business processing flows.

[0048] In this embodiment, after converting the natural language description text of the business processing flow of various services into a graph language description text, the graph language description text can be parsed to extract the upstream nodes, downstream nodes, and annotation information of each business processing node. Then, based on the extracted content, a directed graph set corresponding to the business processing flow of various services is constructed using an adjacency list, an inverse adjacency list, and a node attribute mapping table. Thus, the directed graph set includes the upstream node set, the downstream node set, and the annotation information of each business processing node.

[0049] The method provided in this application can perform syntax conversion and node annotation on natural language description texts of business processing flows for various services, obtaining graphical language description texts of various business processing flows, and then generating a directed graph set based on the graphical language description texts. This allows each business processing node involved in the pending business to be executed by referring to the directed graph set during business processing, avoiding omissions of business processing nodes and ensuring the completeness and orderliness of business processing.

[0050] The embodiments described above illustrate a scheme for determining and extracting the business execution path corresponding to the current business processing node from a directed graph set. In another embodiment of this application, downstream nodes of the current business processing node can be incorporated into the business execution path. For example, the "determining the business execution path of the current business processing node" mentioned above can include, for instance, […]. Figure 5 The steps shown are as follows: Step 301: Determine the downstream nodes of the current business processing node based on the execution order of the current business processing node.

[0051] In this embodiment of the application, the downstream node of the current business processing node can be determined based on the execution order among the business processing nodes in the directed graph set; alternatively, the downstream node set of the current business processing node can be obtained from the directed graph set to determine the downstream node of the current business processing node.

[0052] Step 302: Construct a business execution path based on the current business processing node, the historical business processing nodes that were executed before the current business processing node, and the downstream nodes of the current business processing node.

[0053] In this embodiment, the business execution path corresponding to the current business processing node can be determined based on the current business processing node, the historical business processing nodes executed before the current business processing node, and the downstream nodes of the current business processing node, as well as the execution order among the business processing nodes, and the business execution path can be extracted from the directed graph set.

[0054] In one possible implementation, when obtaining the downstream nodes of the current business processing node, two layers of downstream nodes can be obtained: the downstream nodes directly connected to the current business node, and the downstream nodes connected to the downstream nodes of the current business node.

[0055] For example, the current business processing node is "Can I troubleshoot this myself?", and the corresponding historical business processing nodes include "Inquire about the fault phenomenon", "Inquire about the product model", "Inquire about the product category", and "Determine the service type". The corresponding downstream nodes include "Are you willing to self-service?" and "Enter the appointment information collection", and the downstream nodes of the downstream nodes include "Send self-service guidance SMS" and "Inquire about the appointment time". Therefore, by extracting the above business processing nodes and the connection relationships between each business processing node from the directed graph set, the business execution path of the current business processing node can be obtained.

[0056] The method provided in this application incorporates downstream nodes or multiple layers of downstream nodes of the current business processing node into the business execution path, so that the business processing model can identify in advance the business processing logic that may be executed subsequently, the business data to be collected, etc. For situations where a user provides multiple pieces of business data at once at the current business processing node, such as during a "Can I troubleshoot this myself?" session where the user answers "No, I can't troubleshoot this myself, I'd like to schedule a manual repair appointment for tomorrow afternoon," the server 20 can store multiple pieces of business data to avoid repeated queries at subsequent business processing nodes. For example, it can avoid repeatedly executing "ask for appointment time," thereby improving business processing efficiency.

[0057] In one embodiment, adjacent nodes with similar business execution logic to the current business processing node can also be incorporated into the business execution path. For example, the "determining the business execution path of the current business processing node" mentioned above may include, for instance,... Figure 6 The steps shown are as follows: Step 401: For each historical business processing node, determine the adjacent nodes of the historical business processing node.

[0058] Among them, the parent node of the adjacent node is the same as that of the historical business processing node.

[0059] In this embodiment of the application, when tracing back through the parent node to a historical business processing node that was executed before the current business processing node, adjacent nodes sharing the same parent node as the historical business processing node can be identified. For example, when the current business processing node is "whether self-service exclusion is possible", its historical business node "inquire about product category" has an adjacent node "determine installation category".

[0060] Step 402: Construct a business execution path based on the current business processing node, historical business processing nodes executed before the current business processing node, downstream nodes, and adjacent nodes.

[0061] In this embodiment, the business execution path corresponding to the current business processing node can be determined based on the current business processing node, the historical business processing nodes executed before the current business processing node, the downstream nodes of the current business processing node, the branches where the adjacent nodes of each historical business processing node are located, and the execution order between each business processing node, and the business execution path can be extracted from the directed graph set.

[0062] In one possible implementation, the branches containing adjacent nodes of the current business processing node's historical business processing nodes can be incorporated into the current business processing node's business execution path. That is, the bypasses of the main path containing the current business processing node can be incorporated into the current business processing node's business execution path. Here, the main path refers to the business execution path constructed by the current business processing node and historical business processing nodes executed before the current business processing node. Specifically, after determining all adjacent nodes of each historical business processing node, for each adjacent node, the business processing nodes in the branches containing those adjacent nodes are designated as bypass nodes. Finally, based on the current business processing node, historical business processing nodes, downstream nodes, adjacent nodes, and bypass nodes, the business execution path corresponding to the current business processing node is determined according to the execution order among the nodes. For example, the branch containing the historical business node "Inquire about product category" and the adjacent node "Determine installation category" could be: Determine installation category – Inquire whether the range hood has arrived – Inquire whether the arrival time is known – Enter appointment information collection. Here, "Inquire whether the range hood has arrived," "Inquire whether the arrival time is known," and "Enter appointment information collection" are bypass nodes.

[0063] The method provided in this application embodiment can incorporate adjacent nodes or branches containing adjacent nodes with similar business execution logic into the business execution path. This allows the subsequent business processing model to quickly determine whether the business execution logic of semantically similar adjacent nodes matches the user's new intent when it identifies a conflicting business request intent expressed by the user at the current business processing node. If so, it can quickly redirect, avoiding resetting the entire business processing flow. In this scenario, the business processing result of the current business processing node includes the user's semantics expressed during the business processing at the current business processing node, while the business parsing of the business processing model can include, for example... Figure 7 The steps shown are as follows: Step 501: Match the user semantics with the business processing logic of the current business processing node, the business processing logic of the historical business processing node, and the business processing logic of the downstream node.

[0064] In this embodiment, if a user dialogue exists during the execution of the current business processing node, semantic understanding can be performed on the user dialogue to obtain user semantics. Then, the user semantics are matched with the business processing logic of the current business processing node, the business processing logic of historical business processing nodes, and the business processing logic of downstream nodes to determine whether a change in the user's business request intent has occurred.

[0065] Step 502: If the user semantics do not match the business processing logic of the current business processing node, the business processing logic of the historical business processing node, and the business processing logic of the downstream node, match the user semantics with the business processing logic of the adjacent node, and determine the matched adjacent node as the next business processing node of the current business processing node.

[0066] In this embodiment, if no match is found, the user's intent to change the business request is determined. For example, during the current business processing node's "Inquire about product model" session, after initiating the session "What model of range hood do you want to repair?", the user's dialogue is obtained: "I misspoke before; I wanted to schedule the installation of a newly purchased range hood, not its repair." Semantic understanding of the user's dialogue is performed to obtain user semantics. This user semantics is then matched with the business processing logic of the current business processing node, historical business processing nodes, and downstream nodes to determine that the user's intent to change the business request is product installation.

[0067] Therefore, the user semantics can be matched with the business processing logic of adjacent nodes in the business execution path. If there is an adjacent node whose business processing logic matches the user semantics, then that adjacent node is taken as the next business processing node after the current business processing node, and the business processing flow continues. For example, if the changed user semantics "product installation is required" is matched with the business execution logic of the adjacent node "determine the installation category", then "determine the installation category" is taken as the next business processing node, and the business processing flow continues.

[0068] In one possible implementation, if the next business processing node is determined to be an adjacent node, business data can be populated into the business processing nodes of the branch containing the adjacent node based on the acquired business data. This avoids repeatedly executing the business processing logic of some business processing nodes and improves business processing efficiency. For example, Figure 8As shown, if it is determined that the user's intention changes at the current business processing node "Whether self-service exclusion is possible", and the next business processing node is determined to be "Determine installation category", then business data can be populated into the branch where "Determine installation category" is located (the historical unselected bypass in the figure) based on the already obtained product category (range hood) and product model "CXW-260", avoiding repeatedly asking the user about the product category and product model. In the figure, Partial Future 1: Willing to self-service, Partial Future 2: Direct appointment, and appointment time represent the multiple downstream nodes of the current node "Whether self-service exclusion is possible".

[0069] The method provided in this application can incorporate adjacent nodes or branches containing adjacent nodes with similar business execution logic into the business execution path. When a user's business request intent changes, the business execution logic of semantically similar adjacent nodes can be quickly determined, and a rapid jump can be made if the new intent of the user is met, avoiding the need to reset the entire business processing flow and effectively improving business processing efficiency.

[0070] The embodiments described above illustrate that, in addition to the next business processing node, the business parsing results output by the business processing model may also include response text. To prevent the server from replying to users with risky text via the terminal platform, risk verification of the response text output by the business processing model is required before jumping to the next business processing node. This embodiment may include, for example... Figure 9 The steps shown are as follows: Step 601: Match the reply text with the risk keywords in the preset risk keyword set.

[0071] In this embodiment, a set of risk keywords and risk handling rules can be pre-constructed based on the experience of professionals and stored locally on server 20. The risk handling rules include the correspondence between different risk keywords and standard response texts. Risk keywords may involve costs, transfer of communication platforms, or breach of promises (e.g., the fault you described can be repaired). Correspondingly, standard response texts may include phrases such as "The specific cost will be determined by the repair technician based on the charging standard after on-site inspection" or "The repair result is subject to the on-site judgment of the repair technician."

[0072] After the business processing model outputs the response text, the server 20 matches the response text with the risk keywords in the preset risk keyword set to determine whether the response text contains any violations.

[0073] Step 602: If there are risk keywords that match the response text, determine the target response text based on the matched risk keywords and risk handling rule information.

[0074] In this embodiment, if a risk keyword matching the response text exists in the preset risk keyword set, it indicates that the response text output by the business processing model violates regulations and cannot be displayed to the user. Therefore, based on the risk keyword matching the response text, a query is performed in the risk processing rule information to determine the standard response text corresponding to the risk keyword, thus obtaining the target response text.

[0075] Step 603: Replace the response text based on the target response text.

[0076] In this embodiment, the reply text output by the business processing model is deleted, and the target reply text is displayed to the user through the terminal device 10.

[0077] The method provided in this application embodiment can detect non-compliant response texts output by the business processing model through a preset set of risk keywords, and output standard response texts through risk processing rule information to avoid generating non-compliant responses.

[0078] In one embodiment, the directed graph set can also be rendered as a visual flowchart and displayed on the terminal device 10, and the business execution status can be synchronized to facilitate users' understanding of the business execution progress. That is, the business processing method can also include the following steps: after the business processing logic of the current business processing node is completed, the execution status of the current business processing node is updated, and the execution status is sent to the terminal device 10 to instruct the terminal device 10 to update the visualization style information of the current business node in the visual directed graph based on the execution status.

[0079] Among them, the visualized directed graph is a visualized flowchart of a set of directed graphs corresponding to various business operations.

[0080] In this embodiment of the application, after the construction of the directed graph set is completed, the directed graph set can be rendered as follows: Figure 3 The visualized directed graph is shown and sent to terminal device 10. After a user initiates a service request through terminal device 10, terminal device 10 can respond to the service request by displaying the visualized directed graph to the user.

[0081] During the execution of pending business processes, server 20 updates the execution status of the current business processing node based on the business processing result after completing the business processing logic of the current business processing node. For example, it updates the execution status of the current business processing node to "execution successful" or "execution failed". Then, it sends the execution status of the current business processing node to terminal device 10, instructing terminal device 10 to update the visualization style information of the current business node in the visualized directed graph according to the style information table based on the execution status. The style information table can include the correspondence between different execution statuses and visualization style information (highlight, grayscale, line drawing, etc.). For example, if the execution status of the current business processing node is "execution successful", terminal device 10 is instructed to highlight the current business node in the visualized directed graph; if the execution status of the current business processing node is "execution failed", terminal device 10 is instructed to grayscale the current business node in the visualized directed graph.

[0082] The method provided in this application can render a directed graph set as a visual flowchart displayed on a terminal device, and update the node styles in the visual directed graph displayed on the terminal device based on the execution progress of the business being processed by the server, so as to synchronize the business execution progress of the terminal device and the server. It can also facilitate professionals to quickly and effectively perform fault node location, business review, and other processing.

[0083] In one embodiment, the business processing logic of a business processing node may include collecting business data based on slot information, so that the business processing result may include the collected business data. In another embodiment of this application, the collected business data may be verified, which may include, for example... Figure 10 The steps shown are as follows: Step 701: Verify the business data of the current business processing node and obtain the verification result.

[0084] In this embodiment, for business processing nodes that need to acquire business data, a business data verification process can be set in their business processing logic. For the current business processing node, if business data is collected by executing the corresponding business processing logic, the business data can be verified for type, format, etc., based on the pre-set business data verification process to obtain the verification result. For example, the format of a user-inputted mobile phone number can be verified to obtain a verification result (e.g., an error in the number of digits). Step 702: Input the business execution path, business processing result, and verification result into the business processing model for business parsing.

[0085] In this embodiment, the business execution path, business processing result, and business data verification result corresponding to the current business processing node can be input together as a prompt into the business processing model for business parsing. This allows the business processing model to determine that the next business processing node is still the current business processing node, and to re-collect the business data, even if the business data verification result indicates an error in business data collection. This avoids continuing to execute the business processing logic of other business processing nodes when business data is incorrect, ensuring the reliability and accuracy of business processing.

[0086] In one embodiment, such as Figure 1 As shown, a business processing system is provided, which includes a terminal device 10 and a server 20. The server 20 is used to backtrack historical business processing nodes in a directed graph set based on the current business processing node of the business to be processed, and determine the business execution path of the current business processing node. The directed graph set includes business processing nodes of various businesses and the execution order between the business processing nodes. The server 20 executes the business processing logic of the current business processing node to obtain the business processing result of the current business processing node. The server 20 inputs the business execution path and the business processing result into the business processing model for business parsing to obtain the next business processing node of the business to be processed, and jumps to the next business processing node to execute the corresponding business processing logic, until the next business processing node is the end node in the directed graph set.

[0087] This application provides a service processing system where the server can pre-construct a directed graph set based on the service processing nodes of various services and the execution order between nodes. When executing a service to be processed, the server determines the service execution path of the current service processing node by tracing back through historical service processing nodes in the directed graph set, and executes the service processing logic of the current service processing node to obtain the service processing result. The service execution path and service processing result are input into the service processing model for service parsing to obtain the next service processing node for the service to be processed, and the server jumps to the next service processing node to execute the corresponding service processing logic, until the next service processing node is the end node in the directed graph set. The method provided in this application can refer to the pre-constructed directed graph set of service processing nodes and the execution order to execute each service processing node involved in the service to be processed, avoiding omission of service processing nodes and ensuring the integrity and orderliness of service processing. Furthermore, this application embodiment uses the business execution path and business processing result of the current business processing node as model input for business analysis, avoiding inputting the entire business processing flow into the large language model for analysis. While ensuring the richness of the data input to the model, it effectively reduces the amount of data, avoids affecting the attention of the large language model, and improves the efficiency and accuracy of business parsing based on the large language model.

[0088] It should be noted that although the operations of the method of the present invention are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all of the operations shown must be performed to achieve the desired result. On the contrary, the steps depicted in the flowchart may be performed in a different order. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0089] The following is for reference. Figure 11 It shows a schematic diagram of the structure of a computer system 1100 suitable for implementing terminal devices or servers in the embodiments of this application.

[0090] like Figure 11 As shown, the computer system 1100 includes a central processing unit (CPU) 1101, which can perform various appropriate actions and processes based on programs stored in read-only memory (ROM) 1102 or programs loaded from storage section 1108 into random access memory (RAM) 1103. The RAM 1103 also stores various programs and data required for the operation of the system 1100. The CPU 1101, ROM 1102, and RAM 1103 are interconnected via a bus 1104. An input / output (I / O) interface 1105 is also connected to the bus 1104.

[0091] The following components are connected to I / O interface 1105: an input section 1106 including a keyboard, mouse, etc.; an output section 507 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 1108 including a hard disk, etc.; and a communication section 1109 including a network interface card such as a LAN card, modem, etc. The communication section 1109 performs communication processing via a network such as the Internet. A drive 1110 is also connected to I / O interface 1105 as needed. Removable media 1111, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 1110 as needed so that computer programs read from them can be installed into storage section 1108 as needed.

[0092] In particular, according to embodiments of this disclosure, the above references Figure 11 The described process can be implemented as a computer software program. For example, embodiments of this disclosure include a computer program product comprising a computer program tangibly embodied on a machine-readable medium, the computer program containing instructions for performing... Figure 11 The program code for the method. In such an embodiment, the computer program can be downloaded and installed from a network via communication section 1109, and / or installed from removable media 1111.

[0093] It should be noted that the computer-readable medium shown in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media can also be any computer-readable medium other than computer-readable storage media, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0094] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0095] The units or modules described in the embodiments of this application can be implemented in software or hardware. The described units or modules can also be located in a processor. The names of these units or modules do not, in certain circumstances, constitute a limitation on the unit or module itself.

[0096] On the other hand, this application also provides a computer-readable storage medium, which may be included in the computer device described in the above embodiments, or may exist independently and not assembled into the computer device. The aforementioned computer-readable storage medium stores one or more programs that, when used by one or more processors, execute the methods described in this application. For example, it may execute... Figure 2 The steps of the method shown are as follows.

[0097] This application provides a computer program product including instructions that, when executed, cause the method described in this application to be performed. For example, it can execute... Figure 2 The steps of the method shown are as follows.

[0098] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0099] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A business processing method, characterized in that, The method includes: Based on the current business processing node of the pending business, the historical business processing nodes are traced back in the directed graph set to determine the business execution path of the current business processing node; the directed graph set includes business processing nodes of various businesses and the execution order between each business processing node; Execute the business processing logic of the current business processing node to obtain the business processing result of the current business processing node; The business execution path and the business processing result are input into the business processing model for business parsing to obtain the next business processing node of the business to be processed, and the corresponding business processing logic is executed at the next business processing node until the next business processing node is the end node in the directed graph set.

2. The method according to claim 1, characterized in that, The process of constructing the directed graph set includes: The system obtains natural language description texts of business processing flows for various business operations, performs grammatical transformation and node annotation on the natural language description texts based on a preset chart language, and obtains chart language description texts of various business processing flows; the node annotations include processing logic annotations, attribute annotations, and connection relationship annotations for each business processing node. The chart language description text is parsed to generate a set of directed graphs corresponding to various business processing flows; the set of directed graphs includes the upstream node set, downstream node set, and annotation information of each business processing node.

3. The method according to claim 1, characterized in that, Determining the business execution path of the current business processing node includes: The downstream node of the current business processing node is determined based on the execution order of the current business processing node; The business execution path is constructed based on the current business processing node, the historical business processing nodes executed before the current business processing node, and the downstream nodes of the current business processing node.

4. The method according to claim 3, characterized in that, Determining the business execution path of the current business processing node includes: For each of the aforementioned historical service processing nodes, the adjacent nodes of the historical service processing node are determined; the adjacent nodes are the same as the parent node of the historical service processing node. The business execution path is constructed based on the current business processing node, the historical business processing nodes executed before the current business processing node, the downstream nodes, and the adjacent nodes.

5. The method according to any one of claims 1-4, characterized in that, The business processing result includes the user semantics expressed by the user during the business processing process at the current business processing node, and the business parsing of the business processing model includes: The user semantics are matched with the business processing logic of the current business processing node, the business processing logic of the historical business processing node, and the business processing logic of the downstream node, respectively. If the user semantics do not match the business processing logic of the current business processing node, the business processing logic of the historical business processing node, and the business processing logic of the downstream node, the user semantics are matched with the business processing logic of the adjacent node, and the matched adjacent node is determined as the next business processing node of the current business processing node.

6. The method according to claim 3, characterized in that, Determining the business execution path of the current business processing node includes: For each of the historical service processing nodes, the adjacent nodes and bypass nodes of the historical service processing node are determined; the adjacent node is the same as the parent node of the historical service processing node, and the bypass node is the service processing node in the branch where the adjacent node is located. The business execution path is constructed based on the current business processing node, the historical business processing nodes executed before the current business processing node, the downstream node, the adjacent node, and the bypass node.

7. The method according to claim 1, characterized in that, The business parsing results output by the business processing model also include response text. Before jumping to the next business processing node, the method further includes: The response text is matched with risk keywords in a preset risk keyword set; If a risk keyword matching the response text exists, the target response text is determined based on the matching risk keyword and risk handling rule information; the risk handling rule information includes the correspondence between different risk keywords and standard response texts. Replace the response text with the target response text.

8. The method according to claim 1, characterized in that, The method further includes: After the business processing logic of the current business processing node is completed, the execution status of the current business processing node is updated, and the execution status is sent to the terminal device to instruct the terminal device to update the visualization style information of the current business node in the visualization directed graph based on the execution status; the visualization directed graph is a visualization flowchart of the set of directed graphs corresponding to the various businesses.

9. The method according to claim 1, characterized in that, The business processing result includes the collected business data, and the method further includes: The business data of the current business processing node is verified to obtain the verification result. The business execution path, the business processing result, and the verification result are input into the business processing model for business parsing.

10. A business processing system, characterized in that, It includes a terminal device and a server, wherein the server is used to implement the business processing method according to any one of claims 1-9.

Images