群智软件开发中的人工智能  

Artificial Intelligence in Crowd Software Development

作　　者：蔡维德 张力[1] 胡舒风 CAI Wei-de;ZHANG Li;HU Shu-feng(Digital Society and Blockchain Laboratory,Beihang University,Beijing 100191,China)

机构地区：[1]北京航空航天大学数字社会与区块链实验室,北京100191

出　　处：《贵阳学院学报（社会科学版）》2021年第5期42-50,62,共10页Journal of Guiyang University:Social Sciences

基　　金：中国科技部项目“现代服务可信交易理论与技术研究”(项目编号:2018YFB1402700);北京航空航天大学软件环境国家重点实验室、国家973计划项目(项目编号:2013CB329601);国家自然科学基金项目“群智化软件开发的追踪和度量”(项目编号:61690202)。

摘　　要：群智软件已经存在和发展了多年。这些年来,群智软件有了新的进展和变化。群智的初衷是降低软件开发的成本。然而,群智方式在软件开发中遇到了困难,于是转向了软件测试。群智软件测试(CST)已经有一些成功的案例。它具有许多优点,例如减少了软件测试的成本和时间。分析群智在软件开发中面临的问题,可以得出群智更适合软件测试的结论。进一步分析近年来在CST平台中出现的机器学习(ML)方法能够发现,这些方法可以减少群智人员的工作量。总的来说,群智软件测试面临很多挑战,相关平台也提供了很多基于人工智能的解决方案。Over the years since its existence,Crowd Software has witnessed a lot of new advances and changes.The initial purpose of the Crowd Software was to reduce the cost of software development.However,it has encountered difficulties in software development and so shifted to software testing.There are already some successful cases for Crowd Software Testing(CST)for many of its advantages,including the reduction of cost and time of software testing.After analyzing the problems it has encountered in software development,this paper concludes that it is more suitable for software testing.This paper then makes an analysis of machine learning(ML)methods that have emerged in the CST platform in recent years,which can reduce the workload of Crowd Software personnel.Generally speaking,there are still many challenges for CST while these platforms provide many AI-based solutions.

关 键 词：群智 机器学习 软件测试 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]