Topic

AI Checklist: Qunar's Efficiency Practices in the Testing Domain with AIGC

Introduction: With the wide application of ChatGPT, a large number of natural language processing tasks can be automated. In the testing process, checklist writing is the process of combing, translating and understanding the natural language of requirement documents by qa roles, so it is possible to automate the checklist generation of requirement documents with the help of large model AI, thus reducing labor costs. And for self-testing self-testing projects to generate chekclist to assist the development of self-testing, improve the efficiency and quality of project development and testing. Outline: 1.Introduction to the pain points in the testing domain of GoWhere 2.AI checklist project design program a)Industry reference samples b)Solutions explored by Qunar c)System architecture i.Requirement document preprocessing module ii.Requirement Point Analysis Module iii.AIGC Infrastructure Module iv.Checklist conversion module d) Sample Demonstration 3.Effectiveness evaluation program for auto-generated cases a)Introduction of effect evaluation indexes b)Comparison of effect evaluation schemes c) Introduction of automatic effect evaluation scheme based on embedding. 4. Achievements and follow-up plans a)Current effect b)Future plan