SBIR Phase II: Artificial Intelligence Powered Software Testing

小型企业创新研究（SBIR）第二阶段项目的更广泛/商业影响通过使人工智能（AI）自动化测试而无需编码或经验丰富的编码人员，降低了软件质量保证（SQA）端到端测试的成本和速度。 随着创新型高增长的软件即服务（SaaS）公司以更高的信心和更少的软件缺陷更快地进入市场，行业将通过节省时间和金钱而在经济上受益。SBIR第二阶段项目将构建一个人工智能解决方案，尽管该解决方案由经验不足的软件工程师使用，但将允许软件公司以最少的用户交互来识别软件缺陷。实时和引导过程直接从Web浏览器收集信息，处理传统的和未解决的测试自动化问题，如软件测试设计，自动化，覆盖率和维护。人工智能解决方案将通过执行两项主要任务使SQA高效：模拟实时用户对Web应用程序的探索和识别意外行为。该架构使AI代理能够自我学习并与应用程序交互，从而改进每次观察。人工智能学习周期在系统内实现了彻底的沟通，因为它根据自己的知识分析产生的效果来传达应用特定行动的请求。第一阶段的研究证明，该架构可以升级到商业版本，为希望提高产品软件质量并更快上市的客户提供价值。第二阶段的预期技术成果将加强对意外软件行为的分类，优化数据分析时间，并缩短学习周期。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The broader/commercial impact of the Small Business Innovation Research (SBIR) Phase II project reduce the cost and speed of software quality assurance (SQA) end-to-end testing by enabling artificial intelligence (AI) to automate tests without the need for coding or highly experienced coders. As innovative high-growth Software as a Service (SaaS) companies go to market faster with more confidence and fewer software defects, industries will benefit economically by saving time and money. This SBIR Phase II project will build an AI solution which, although used by less experienced software engineers, will allow software companies to identify software defects with minimal user interactions. The real-time and guided process gathers information directly from the web browsers, handling traditional and unresolved problems with test automation such as software test design, automation, coverage, and maintenance. The AI solution will make SQA highly efficient by performing two major tasks: simulating real-time users' exploration of web applications and identifying unexpected behaviors. The architecture enables AI agents to self-learn and interact with the application, improving on each observation. The AI learning cycle implements thorough communication within the system as it communicates requests to apply specific actions based on its own knowledge analyzing the resulting effect. Phase I research proved that the architecture can be upgraded to a commercial version, providing value to customers looking to improve software quality in their products and go to market faster. The anticipated technical results in Phase II will enhance the categorization of unexpected software behaviors, optimize the data analysis time, and reduce the learning cycle.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.