CN-122022296-A - Service processing method and device

CN122022296ACN 122022296 ACN122022296 ACN 122022296ACN-122022296-A

Abstract

The invention discloses a service processing method and device. The method comprises the steps of obtaining a service demand description text, determining a service operation task corresponding to the service demand description text, wherein the service operation task comprises an atomic operation, a standard step flow or a subtask to be executed, obtaining metadata of the service operation task, determining execution information corresponding to the service operation task according to the metadata, wherein the execution information comprises an execution end and an execution time sequence, and completing the service operation task according to the execution information. The method reduces the redundant planning links of the high-frequency simple demands, realizes the optimization of the quick response and the resource consumption of the high-frequency simple demands, solves the problems of task matching failure and automatic business process stagnation caused by a scheme without automatic bottom and complex demands disassembly when facing undefined complex compound tasks, and further improves the rationality of the resource allocation of an executing end through accurately determining the executing information so as to further improve the efficiency and the resource utilization rate of the business automatic execution.

Inventors

XIAO CHENGHONG
LI ZEBIN
Zhong Yousi
QIU JIAN
HE HUIMIN
ZHOU YUN
REN XIAOJUN
LIU CHUNLIN
CHU RUI
CHEN GUO
Kou Chenhui

Assignees

中移信息技术有限公司
中国移动通信集团有限公司

Dates

Publication Date: 20260512
Application Date: 20260121

Claims (10)

1. A method for processing a service, comprising: Acquiring a service demand description text; Determining a business operation task corresponding to the business demand description text, wherein the business operation task comprises an atomic operation, a standard step flow or a subtask to be executed; acquiring metadata of the business operation task, and determining execution information corresponding to the business operation task according to the metadata, wherein the execution information comprises an execution end and an execution time sequence; And completing the business operation task according to the execution information.
2. The method of claim 1, wherein the determining the business operation task corresponding to the business requirement description text comprises: Matching the service demand description text based on a preset regular expression, and determining whether a matching result meets a preset first matching condition; If yes, determining a word vector of the service demand description text, and determining an operation hit result according to the word vector and a pre-constructed atomic operation library; Under the condition that the operation hit result is that an atomic operation is hit, determining a business operation task according to the hit atomic operation; If the operation hit result is not met or the operation hit result is not hit atomic operation, determining whether a standard step flow matched with the service demand description text exists or not by querying a pre-built standard flow knowledge graph, and determining the standard step flow as a service operation task if the standard step flow exists; And under the condition that the standard step flow does not exist, determining a subtask to be executed according to time steps by a plurality of agents, and determining the subtask to be executed as a business operation task.
3. The method according to claim 2, wherein the step of constructing the normalized flow knowledge graph includes: Extracting a triplet set from a text corresponding to at least one standardized flow according to preset mode information, wherein the mode information comprises entity type constraint information, entity relationship constraint information and entity key attribute constraint information; Determining a first embedded vector of each entity and a second embedded vector of a corresponding relation of the entity according to a preset embedding algorithm, a triplet related to the entity and a trained graph rolling network, wherein the graph rolling network is determined through iterative training of a positive sample set formed by the acquired real triplet and a negative sample set formed by a false triplet constructed according to a preset construction strategy; For each triplet in the triplet set, determining an evaluation function value of the triplet according to a first embedded vector of a head entity, a first embedded vector of a tail entity and a second embedded vector of a relation by combining a preset evaluation function; and constructing a standardized flow knowledge graph according to the target triples of which the evaluation function values meet the quality evaluation conditions.
4. The method of claim 2, wherein the plurality of agents includes a master agent, an execution agent, an evaluation agent, and a memory agent; correspondingly, the determining, by a plurality of agents, the subtasks to be executed according to the time steps includes: Aiming at the first time step, determining a subtask sequence to be executed of the current time step according to the service demand description text and the interactive interface state information of the current time step by the main agent; Aiming at each time step except the first time step, determining a subtask sequence to be executed of the current time step according to the description text of the service requirement, the state information of the interactive interface of the current time step, the subtask sequence to be executed of the previous time step, the completed subtask sequence, the execution action information, the evaluation feedback and the accumulated memory by the main intelligent agent; and determining the subtasks to be executed corresponding to the current time step according to the service demand description text, the subtask sequence to be executed and the interactive interface state information of the current time step, and the evaluation feedback and the accumulated memory of the previous time step by the execution agent.
5. The method of claim 4, wherein the completed sub-task sequence for each time step is determined by the master agent by identifying sub-tasks in the sequence of sub-tasks to be performed for the time step that have been successfully performed; the execution action information of each time step is determined by the execution agent according to the subtasks to be executed, and the execution action information comprises the subtasks to be executed, target actions and expected action effects; The evaluation feedback of each time step is determined by the evaluation agent according to the service demand description text, the interactive interface state information before the target action is executed, the interactive interface state information after the target action is executed and the execution action information; The accumulated memory of each time step is determined by the memory agent based on the interactive interface state information of the last time step, the interactive interface state information of the current time step, and the accumulated memory of the last time step.
6. The method according to claim 1, wherein, in the case that the business operation task is an atomic operation or a subtask to be executed, the determining, according to the metadata, execution information corresponding to the business operation task includes: Determining a target attribute label matched with the business operation task through keyword recognition or flow template matching according to a pre-constructed task attribute library and the metadata, wherein the task attribute library comprises a mapping table between operation task types and attribute labels, and the attribute labels comprise a local adaptation attribute, a cloud adaptation attribute, a high real-time attribute and a low real-time attribute; Under the condition that the target attribute tag comprises a local adaptation attribute, determining an execution end of the business operation task as a local terminal; under the condition that the target attribute tag comprises cloud adaptation attributes, determining an execution end of the business operation task according to current environment perception data; When the target attribute tag comprises high real-time attribute and/or the business operation task is determined to have a front-end process chain dependency relationship, determining the execution time sequence of the business operation task as synchronous execution; And when the target attribute tag comprises low real-time attribute and the business operation task is determined to have no pre-process chain dependency relationship, determining the execution time sequence of the business operation task to be asynchronous execution.
7. The method according to claim 1, wherein, in the case that the business operation task is a standard step flow, the determining, according to the metadata, execution information corresponding to the business operation task includes: Determining a target attribute tag sequence matched with the standard step flow through keyword recognition or flow template matching according to a pre-constructed task attribute library and the metadata, wherein one target attribute tag corresponds to one standard step; And determining an execution end and an execution time sequence of the corresponding standard step according to each target attribute tag in the target attribute tag sequence.
8. The method of claim 1, wherein said completing said business operations task based on said execution information comprises: Distributing the business operation task to a corresponding terminal according to an execution end in the execution information; Determining a task scheduling strategy corresponding to the service operation task by the terminal according to the execution time sequence in the execution information, and executing the service operation task according to the task scheduling strategy; And under the condition that the business operation task is a subtask to be executed, returning to execute the relevant steps of determining the business operation task corresponding to the business requirement description text again until the business process ending condition is reached.
9. The method as recited in claim 1, further comprising: According to the preset extraction frequency, acquiring log data in the business process advancing process; Updating an atomic operation library according to a preset first statistical index, a first triggering condition and a first updating logic by combining the log data, wherein the first statistical index comprises a miss frequency, an atomic operation multiplexing rate, a complex operation disassembling complexity and an atomic operation execution success rate; And updating the standardized flow knowledge graph by combining the log data according to a second preset statistical index, a second trigger condition and a second update logic, wherein the second statistical index comprises task repetition frequency, multi-agent planning complexity and workflow stability.
10. A service processing apparatus, comprising: The acquisition module is used for acquiring the service demand description text; The task determining module is used for determining a business operation task corresponding to the business requirement description text, wherein the business operation task comprises an atomic operation, a standard step flow or a subtask to be executed; the execution information determining module is used for acquiring metadata of the service operation task and determining execution information corresponding to the service operation task according to the metadata, wherein the execution information comprises an execution end and an execution time sequence; And the task execution module is used for completing the business operation task according to the execution information.

Description

Service processing method and device Technical Field The present invention relates to the field of computer processing technologies, and in particular, to a service processing method and apparatus. Background In the process of enterprise digital transformation, service automation execution becomes a core requirement for cost reduction and efficiency improvement. Robot flow automation is a technique that can automatically perform repetitive tasks with high efficiency and high accuracy. In practical application, the robot process automation remarkably improves the service processing efficiency by automatically executing the standardized process. Otherwise, the standardized flow provides a standardized input framework for automatic execution of the robot flow, and the standardized flow and the robot flow have natural cooperative association. However, there are many limitations in practical applications in existing business automation processing methods involving standardized processes and robotic process automation. On the one hand, autonomous dynamic programming capability is not sufficient. The existing method can only process the predefined standardized workflow, when the business process matched with the requirement of the interactive object is not searched, the method completely relies on manual filling of the knowledge blank, and no automatic bottom approach exists, and especially under the condition that a user delays the response, the whole process is stopped, and the closed loop completion quality of the business process is seriously affected. On the other hand, the resource allocation of the execution end is unreasonable. The existing method generally optimizes the utilization of remote resources around the priorities and the dependency relationships of subtasks, but does not consider the self characteristics of service operation tasks and the adaptation problem of execution end resources, so that the problems of rough local and remote cooperation and unreasonable resource allocation occur, further resource waste or task blockage is caused, and the service flow processing efficiency is affected. Disclosure of Invention The invention provides a service processing method and a service processing device, which are used for solving the problems of insufficient autonomous dynamic programming capability and unreasonable resource allocation of an execution end of the existing service automation processing method. In a first aspect, an embodiment of the present invention provides a service processing method, where the method includes: Acquiring a service demand description text; Determining a business operation task corresponding to the business demand description text, wherein the business operation task comprises an atomic operation, a standard step flow or a subtask to be executed; acquiring metadata of the business operation task, and determining execution information corresponding to the business operation task according to the metadata, wherein the execution information comprises an execution end and an execution time sequence; And completing the business operation task according to the execution information. In a second aspect, an embodiment of the present invention provides a service processing apparatus, including: The acquisition module is used for acquiring the service demand description text; The task determining module is used for determining a business operation task corresponding to the business requirement description text, wherein the business operation task comprises an atomic operation, a standard step flow or a subtask to be executed; the execution information determining module is used for acquiring metadata of the service operation task and determining execution information corresponding to the service operation task according to the metadata, wherein the execution information comprises an execution end and an execution time sequence; And the task execution module is used for completing the business operation task according to the execution information. According to the technical scheme, the business operation tasks corresponding to the business requirement description text are determined, the business operation tasks comprise atomic operations, standard step flows or subtasks to be executed, metadata of the business operation tasks are obtained, execution information corresponding to the business operation tasks is determined according to the metadata, the execution information comprises an execution end and an execution time sequence, and the business operation tasks are completed according to the execution information. By utilizing the method, the service operation tasks which are accurately adapted to the service demand description text are determined from the three service operation tasks, so that the redundant planning links of the high-frequency simple demands are reduced, the quick response to the high-frequency simple demands and the optimization of resource consumption are rea