CN-121998652-A - Multi-scene intelligent customer service system and method for building in energy industry

CN121998652ACN 121998652 ACN121998652 ACN 121998652ACN-121998652-A

Abstract

The application discloses a multi-scene intelligent customer service system and a multi-scene intelligent customer service method for building in the energy industry, which belong to the technical field of computers. By means of joint analysis of the previewing log and the business data model, conflicts caused by dynamic mismatch of business data or user expression difference can be identified, and root causes and influence ranges of the conflicts can be analyzed. Based on the method, a structured optimization strategy can be generated and iterative verification is carried out, so that the problems of insufficient coupling, conflict recognition lag and optimization dependence trial-and-error of the traditional customer service system in a complex dynamic service environment are solved, and the robustness and deployment efficiency of the intelligent customer service system are improved.

Inventors

GU JIANGHUA
LIU JIAWEN
CHEN XINBAO
MA HAITAO
HU LIPING

Assignees

杭州艾草信息服务有限公司

Dates

Publication Date: 20260508
Application Date: 20260410

Claims (10)

1. The method for building the intelligent multi-scene customer service in the energy industry is characterized by comprising the following steps of: Converting a customer service scene template into a scene flow digital twin body based on service specifications, carrying out structuring and semanteme processing on equipment account, dynamic policies and historical interaction data, and constructing a multi-source service data model which is matched with the scene flow digital twin body; In a dynamic service sandbox, dynamically coupling the scene flow digital twin with real-time service data in the multi-source service data model, and driving a dialogue engine in the sandbox to execute multiple rounds of automatic simulation dialogue to generate a previewing log containing node jump, parameter transfer and external data calling results; Performing joint analysis on the previewing log and the multi-source business data model, identifying dynamic business conflicts caused by dynamic mismatch of business data or user expression difference, and analyzing to obtain root cause category and affected flow node set of the dynamic business conflicts; based on the root cause category and the affected flow node set, generating a structural optimization strategy aiming at the scene flow digital twin body, and iteratively executing the dynamic coupling, the automatic simulation dialogue and the joint analysis until the preview is passed, and outputting the intelligent customer service flow of which the verification is completed.
2. The method of claim 1, wherein the converting customer service scene templates into scene flow digital twins based on the business specifications comprises: Deconstructing the customer service scene template based on the service specification to generate a flow element set, wherein the flow element set comprises atomic service intention and standardized system actions; recombining the flow element set according to the service logic defined in the service specification to construct an initial flow skeleton with a state transition path; And configuring an observation hook for triggering internal state acquisition and exposure during the replay of the dynamic business sandbox to form the scene flow digital twin body for at least part of state nodes of the initial flow skeleton, wherein the observation hook is used for enabling the dynamic business sandbox to capture the detailed execution context of the state nodes.
3. The method of claim 1, wherein the structuring and semantically processing the device ledger, the dynamic policy, and the historical interaction data to construct a multi-source business data model adapted to the scene flow digital twin, comprises: Executing domain-specific structured analysis on the equipment ledger, the dynamic policy and the historical interaction data, extracting and packaging to form a business data object with a version identifier; Based on a predefined business entity map, carrying out semantic merging and conflict resolution on the business data object to generate a standardized business entity set carrying a consistency check label; And configuring a dynamic parameter interface which can be called by the dynamic service sandbox in real time for the entity in the standardized service entity set, and a state evolution rule which can be related to the scene flow digital twin node logic, so as to generate the multi-source service data model with dynamic response and logic constraint capability.
4. The method of claim 1, wherein dynamically coupling the scene flow digital twin with real-time business data in the multi-source business data model in a dynamic business sandbox drives a dialog engine in the sandbox to perform multiple rounds of automated simulation dialog, generating a preview log containing node hops, parameter transfer, and external data call results, comprising: Dynamically binding the real-time service data to corresponding process nodes in the dynamic service sandbox based on a preset data mapping relation between the multi-source service data model and the scene process digital twin, and completing the initialization of a simulation environment; generating a simulated user input sequence set covering a plurality of key dialogue paths in the scene flow digital twin based on a historical interaction mode in the multi-source business data model; driving the dialogue engine to traverse different state transition paths in the scene flow digital twin body according to the simulation user input sequence set, and inquiring and matching dynamic business rules and constraints in the multi-source business data model in real time through a rule interpreter of the dynamic business sandbox in the execution process to generate a multi-round automatic simulation dialogue process; In the automatic simulation dialogue process execution, node execution details triggered by the scene flow digital twin, rule matching and data calling events generated by the rule interpreter, content and analysis results of the simulation user input sequence and path decision information of the dialogue engine are synchronously recorded, and the previewing log is generated through fusion.
5. The method of claim 4, wherein generating a simulated user input sequence set that covers a plurality of critical conversation paths in the scene flow digital twin based on historical interaction patterns in the multi-source business data model comprises: Extracting a user speaking sample and a system response path from historical interaction data of the multi-source business data model, and analyzing to obtain historical dialogue path coverage statistical information; merging the historical dialogue path coverage statistical information with the topological structure characteristics of the scene flow digital twin, and screening out a target key dialogue path set; generating an adapted initial simulation user input sequence set based on the user session sample and node constraints of each path in the target key dialogue path set; And calling the dynamic service sandbox, driving previewing by using the initial simulation user input sequence set, screening and enhancing the initial simulation user input sequence set based on actual path coverage data triggered by previewing, and generating the simulation user input sequence set in an iteration mode.
6. The method of claim 5, wherein the generating an adapted initial set of simulated user input sequences based on the user session sample and node constraints for each path in the set of target critical dialog paths comprises: Aiming at each path in the target key dialogue path set, analyzing service intention constraint and parameter constraint associated with key nodes in each path to generate a node constraint configuration set; Taking the node constraint configuration set as a condition, carrying out multidimensional semantic matching and constraint compliance verification on the user telephone sample, screening out a compliance user telephone sample, carrying out text variation on the compliance user telephone sample on the premise of retaining original intention semantics based on semantic similarity, and generating a diversified telephone candidate set; selecting an adaptation telephone from the diversified telephone candidate set for each key node according to the execution time sequence and logic dependence of the nodes on each path, and carrying out serialization arrangement to generate an initial simulation user input subsequence covering each path; and aggregating the initial simulation user input subsequences generated for all the target key dialogue paths, and performing cross-path redundancy elimination and distribution balance optimization to form the initial simulation user input sequence set.
7. The method of claim 1, wherein the performing a joint analysis of the previewing log and the multi-source business data model to identify dynamic business conflicts due to dynamic mismatch of business data or user expression differences comprises: Extracting parameter verification failure events, rule matching blocking events and user input analysis events which lead to flow execution interruption from the previewing log; associating the parameter verification failure event with the rule matching blocking event to a real-time state of a corresponding service entity in the multi-source service data model, verifying whether the real-time state meets the data requirement of the current node of the scene flow digital twin, and generating a service data dynamic mismatch conflict set; Aiming at the user input analysis event, analyzing the semantic deviation degree between the expression of the user input analysis event and the standard intention preset by the scene flow digital twin, and identifying the analysis event which exceeds a threshold value and actually causes path redundancy as a user expression difference conflict set; And combining the service data dynamic mismatch conflict set and the user expression difference conflict set to form the dynamic service conflict.
8. The method of claim 1, wherein said parsing results in a set of root cause categories and affected flow nodes for the dynamic traffic conflict, comprising: aiming at the dynamic business conflict, locating a flow node in the previewing log, in which the dynamic business conflict first occurs, extracting a data operation context of the flow node, associating the data operation context with a specific business entity and attribute in the multi-source business data model, and determining a direct root node and an abnormal data item; Performing pattern matching on the direct root node and the abnormal data item based on a predefined root classification rule base to determine the root class of the dynamic business conflict, wherein the rules of the root classification rule base are derived from business entity relations and state constraints defined in the multi-source business data model; Analyzing all downstream nodes depending on the abnormal data item or affected by the state transition of the direct root node by taking the direct root node as a starting point along the topological structure of the scene flow digital twin, and determining an affected flow node set by combining with the follow-up execution interruption or path deviation evidence recorded in the preview log.
9. The method of claim 1, wherein the generating a structured optimization strategy for the scene flow digital twin based on the root cause category and the affected flow node set comprises: based on the root cause category, matching a corresponding optimization strategy template from a preset optimization strategy knowledge base, wherein the optimization strategy template comprises node adjustment rules and data adaptation rules; carrying out parameterization binding and logic instantiation on node information and associated abnormal data items in the affected flow node set and rules in the optimization strategy template to generate an optimization instruction set aiming at a specific node; and executing conflict resolution and business logic consistency check on the optimized instruction set, and performing execution priority sorting according to the execution logic of the scene flow digital twin body to generate the structured optimization strategy.
10. An energy industry builds many scenes intelligence customer service system, its characterized in that, the system includes: The construction module is used for converting the customer service scene template into a scene flow digital twin body based on the service specification, carrying out structuring and semanteme processing on equipment account, dynamic policies and historical interaction data, and constructing a multi-source service data model which is adaptive to the scene flow digital twin body; The driving module is used for dynamically coupling the scene flow digital twin body with the real-time service data in the multi-source service data model in the dynamic service sandbox, driving a dialogue engine in the sandbox to execute multiple-time automatic simulation dialogue, and generating a previewing log containing node jump, parameter transfer and external data calling results; The analysis module is used for carrying out joint analysis on the previewing log and the multi-source business data model, identifying dynamic business conflicts caused by dynamic mismatch of business data or user expression difference, and analyzing to obtain root cause categories of the dynamic business conflicts and affected flow node sets; the generation module is used for generating a structural optimization strategy aiming at the scene flow digital twin body based on the root cause category and the affected flow node set, and outputting the intelligent customer service flow after verification by iteratively executing the dynamic coupling, the automatic simulation dialogue and the joint analysis until the previewing passes.

Description

Multi-scene intelligent customer service system and method for building in energy industry Technical Field The application relates to the technical field of computers, in particular to a multi-scene intelligent customer service system and method for building in the energy industry. Background The intelligent customer service system in the energy industry needs to process complex business scenes with equipment account dynamic update, frequent policy adjustment and high diversity of user expression. The current mainstream construction method relies on a static dialogue template and a fixed rule base, and flow logic and service data are in a loose coupling state, so that the perception and adaptation capability to a real-time service environment are lacked. In practical application, the change of service data easily causes the failure of preset flow parameters, causes the jump error of nodes or the abnormality of external calling, and the nonstandard expression of users often causes the intention recognition to deviate from a preset path, thus causing the interruption of conversation. The conventional verification means can only carry out single-point verification based on limited manual test cases, cannot dynamically simulate the coupling effect of multi-source service data and user interaction before deployment, has lag in conflict recognition and fuzzy root cause positioning, and the optimization process is highly dependent on expert experience to repeatedly test errors, and has long period, high cost and difficulty in covering multi-scene boundary conditions. Therefore, how to realize the deep coupling of customer service flows and dynamic service environments, build a previewable, diagnosable and iteratable verification mechanism, recognize conflicts caused by unmatched service data and user expression differences, and drive the automatic optimization of flows, so that the problems to be solved are urgent for improving the robustness and deployment efficiency of intelligent customer service systems in the energy industry. Disclosure of Invention The embodiment of the application provides a multi-scene intelligent customer service system and method for building in the energy industry, and the technical scheme is as follows: In one aspect, a method for building multi-scenario intelligent customer service in an energy industry is provided, and the method comprises the following steps: Converting a customer service scene template into a scene flow digital twin body based on service specifications, carrying out structuring and semanteme processing on equipment account, dynamic policies and historical interaction data, and constructing a multi-source service data model which is matched with the scene flow digital twin body; In a dynamic service sandbox, dynamically coupling the scene flow digital twin with real-time service data in the multi-source service data model, and driving a dialogue engine in the sandbox to execute multiple rounds of automatic simulation dialogue to generate a previewing log containing node jump, parameter transfer and external data calling results; Performing joint analysis on the previewing log and the multi-source business data model, identifying dynamic business conflicts caused by dynamic mismatch of business data or user expression difference, and analyzing to obtain root cause category and affected flow node set of the dynamic business conflicts; based on the root cause category and the affected flow node set, generating a structural optimization strategy aiming at the scene flow digital twin body, and iteratively executing the dynamic coupling, the automatic simulation dialogue and the joint analysis until the preview is passed, and outputting the intelligent customer service flow of which the verification is completed. In one aspect, an energy industry is provided to build a multi-scenario intelligent customer service system, the system comprising: The construction module is used for converting the customer service scene template into a scene flow digital twin body based on the service specification, carrying out structuring and semanteme processing on equipment account, dynamic policies and historical interaction data, and constructing a multi-source service data model which is adaptive to the scene flow digital twin body; The driving module is used for dynamically coupling the scene flow digital twin body with the real-time service data in the multi-source service data model in the dynamic service sandbox, driving a dialogue engine in the sandbox to execute multiple-time automatic simulation dialogue, and generating a previewing log containing node jump, parameter transfer and external data calling results; The analysis module is used for carrying out joint analysis on the previewing log and the multi-source business data model, identifying dynamic business conflicts caused by dynamic mismatch of business data or user expression difference, and analyzing to obtain root cause