CN-121997926-A - Method and device for detecting quality of home broadband assembly maintenance service work orders based on intelligent agent

Abstract

The method for detecting the quality of the home broadband assembly maintenance service work orders based on the intelligent agent is provided. In the method, firstly, a quality inspection request is acquired, quality inspection agents are matched from an agent pool according to the quality inspection request, the quality inspection agents are obtained based on quality inspection rules and an agent element model, and the quality inspection agents comprise quality inspection capability. And secondly, translating the recording file into text data, wherein the recording file is used for recording service calls corresponding to the home broadband assembly maintenance service worksheet. Then, matching the prompting word template according to the quality testing capability of the quality testing agent, and generating prompting words based on the prompting word template and the quality testing rules corresponding to the quality testing agent. And finally, reasoning the text data and the prompt words through the large model to obtain a quality inspection result. By the method, the quality inspection service can be quickly responded to change, and the quality inspection accuracy and the service laminating degree are improved.

Inventors

• GU JIANGUO
• ZHANG XIAOYU
• ZHONG XUEJUN
• DONG ANQI

Assignees

• 杭州东方通信软件技术有限公司

Dates

Publication Date

20260508

Application Date

20251229

Claims (10)

1. 1. A home broadband assembly maintenance service work order quality inspection method based on an intelligent agent is characterized by comprising the following steps: The quality inspection method comprises the steps of acquiring a quality inspection request, and matching a quality inspection agent from an agent pool according to the quality inspection request, wherein the quality inspection agent is obtained based on quality inspection rules and an agent element model, and the quality inspection agent comprises quality inspection capability; The recording file is used for recording service calls corresponding to the home broadband maintenance service worksheet; matching a prompt word template according to the quality testing capability of the quality testing agent, and generating a first prompt word based on the prompt word template and a quality testing rule corresponding to the quality testing agent; and reasoning the text data and the first prompt word through a large model to obtain a first quality inspection result.
2. 2. The method of claim 1, wherein the quality inspection request includes a work order ID, and the converting the sound recording file into text data comprises: Acquiring the recording file based on the work order ID; and calling a translation tool to translate the sound recording file into the text data.
3. 3. The method of claim 1, wherein the quality inspection request includes a home broadband installation service worksheet, the method further comprising, after converting the sound recording file to text data: Acquiring related variables from the home broadband assembly service work order based on the personalized quality inspection item; after the first prompt word is generated based on the prompt word template and the quality inspection rule corresponding to the quality inspection agent, the method further comprises: combining the related variable with the first prompt word to obtain a second prompt word; and reasoning the text data and the second prompt word through the large model to obtain a second quality inspection result.
4. 4. The method according to claim 1, wherein the method further comprises: Acquiring a quality inspection rule; analyzing the quality inspection rule; and generating the quality inspection agent based on the agent element model and in combination with the analysis result of the quality inspection rule.
5. 5. The method of claim 1, wherein the quality inspection rule includes a keyword, the first prompt includes the keyword, the first quality inspection result includes a first matching result of the keyword and a first reasoning result of the large model, and the reasoning the text data and the first prompt through the large model to obtain the first quality inspection result includes: The large model obtains a first matching result of the keyword and a first reasoning result of the large model based on the text data and the first prompt word, wherein the first matching result of the keyword is obtained based on the text data and the keyword in the first prompt word.
6. 6. A method according to claim 3, wherein the quality inspection rule includes a keyword, the second prompt word includes the keyword, the second quality inspection result includes a second matching result of the keyword and a second reasoning result of the large model, and the reasoning of the text data and the second prompt word by the large model to obtain the second quality inspection result includes: The large model obtains a second matching result of the keyword and a second reasoning result of the large model based on the text data and the second prompt word, wherein the second matching result of the keyword is obtained based on the text data and the keyword in the second prompt word.
7. 7. Device is examined to worker simple substance based on intelligent agent, a serial communication port includes: the acquisition module is used for acquiring the quality inspection request; the matching module is used for matching the quality inspection agent from the agent pool according to the quality inspection request; the quality inspection agent is obtained based on quality inspection rules and an agent element model, and comprises quality inspection capability; The translation module is used for translating the sound recording file into text data, wherein the sound recording file is used for recording service calls corresponding to the home broadband maintenance service worksheet; The generation module is used for matching the prompt word template according to the quality inspection capability of the quality inspection intelligent body and generating a first prompt word based on the prompt word template and the quality inspection rule corresponding to the quality inspection intelligent body; and the reasoning module is used for reasoning the text data and the first prompt word through the large model to obtain a first quality inspection result.
8. 8. A computing device comprising a memory and a processor, wherein the memory has instructions stored therein that when executed by the processor cause the method of any of claims 1-6 to be implemented.
9. 9. A computer readable storage medium, on which a computer program is stored which, when being executed by a processor, causes the method according to any of claims 1-6 to be implemented.
10. 10. A computer program product comprising program instructions which, when executed by a computer, cause the computer to perform the method of any of claims 1-6.

Description

Method and device for detecting quality of home broadband assembly maintenance service work orders based on intelligent agent Technical Field The application relates to the technical field of artificial intelligence (ARTIFICIAL INTELLIGENCE, AI), in particular to a home broadband maintenance service work order quality inspection method and device based on an intelligent body. Background At present, in a quality control system of home broadband maintenance service, voice quality inspection is a key link for guaranteeing customer service standardization and maintenance process compliance. The quality inspection service has various types and complex scenes, the prior art is difficult to quickly respond to the change of the quality inspection service, and the quality inspection accuracy and the service fitting degree are difficult to be compatible. For example, the prior art generally adopts methods of fixed keyword rule matching, machine learning quality inspection based on small models, and the like. By adopting a fixed keyword rule matching method, IT is necessary to construct a set of keyword library and matching logic (for example, including negative expressions such as "working cannot be carried out by a person", "no hand" is abused "and the like) in advance by IT engineers or qualification experts according to historical rule violations and service specifications, and set triggering conditions for the keywords. The method has the problems of rule stiffness, difficulty in covering new business talk, spoken language expression, graceful rejection or long-tail violation scenes, and incapability of agilely responding to business changes due to the fact that rule addition or adjustment needs to rely on IT development intervention and the online period is usually several days to several weeks. By adopting the machine learning quality inspection method based on the small model, a large amount of historical voice transfer texts need to be manually marked, the performance of the machine learning quality inspection method is highly dependent on the quality and coverage range of marked data, the model needs to be retrained by service rule change, and once the model is online, the service rule change or the quality inspection dimension is difficult to be quickly adapted. Disclosure of Invention The application provides a home broadband maintenance service work order quality inspection method and device based on an intelligent agent, and solves the technical problems. In a first aspect, a home broadband maintenance service work order quality inspection method based on an intelligent agent is provided. In the method, firstly, a quality inspection request is acquired, quality inspection agents are matched from an agent pool according to the quality inspection request, the quality inspection agents are obtained based on quality inspection rules and an agent element model, and the quality inspection agents comprise quality inspection capability. And secondly, translating the recording file into text data, wherein the recording file is used for recording service calls corresponding to the home broadband assembly maintenance service worksheet. Then, matching the prompt word template according to the quality inspection capability of the quality inspection agent, and generating a first prompt word based on the prompt word template and the quality inspection rule corresponding to the quality inspection agent. And finally, reasoning the text data and the first prompt word through the large model to obtain a first quality inspection result. In one possible implementation, the quality inspection request includes a work order ID, and in converting the sound recording file to text data, first, the sound recording file is acquired based on the work order ID. And secondly, calling a translation tool to translate the sound recording file into text data. In one possible implementation, the quality inspection request includes a home broadband service worksheet. After translating the sound recording file into text data, related variables are obtained from the home broadband service worksheet based on the personalized quality inspection items. After a first prompt word is generated based on the prompt word template and the quality inspection rule corresponding to the quality inspection agent, the related variable is combined with the first prompt word, and a second prompt word is obtained. And reasoning the text data and the second prompt word through the large model to obtain a second quality inspection result. In one possible implementation, the agent-based home broadband maintenance service work order quality inspection method further includes a process of generating quality inspection agents. In this process, quality inspection rules are first acquired. And secondly, analyzing the quality inspection rule. And finally, based on the intelligent agent model, combining the analysis result of the quality inspection rule to generate the qu