Context-Aware Functional Test Generation via Business Logic Extraction and Adaptation
New AI system extracts business logic from requirements to generate functional tests, beating state-of-the-art methods.
A research team from Peking University and Beihang University has introduced LogiDroid, a novel AI-powered approach to automate functional testing for mobile applications. The system addresses two persistent challenges in software testing: extracting complex business logic from unstructured requirements and adapting that logic to diverse graphical user interfaces. LogiDroid operates through a two-stage process—first retrieving and fusing knowledge from relevant cases, then generating context-aware test cases by analyzing both the extracted business logic and real-time GUI environments. This enables the system to understand application semantics and generate complete test cases with verification assertions.
The researchers evaluated LogiDroid using two established datasets (FrUITeR and Lin) covering 28 real-world applications and 190 functional requirements. Experimental results show LogiDroid successfully tested 40% of functional requirements on the FrUITeR dataset—a 48% improvement over state-of-the-art approaches—and 65% on the Lin dataset, representing a 55% improvement. These results demonstrate LogiDroid's ability to bridge the semantic gap between business requirements and GUI implementation, potentially reducing manual testing effort while improving test coverage and accuracy for mobile applications across diverse domains.
- LogiDroid uses two-stage AI to extract business logic from requirements and adapt it to app GUIs
- Tested on 28 real-world apps covering 190 requirements with 40-65% success rates
- Outperforms state-of-the-art methods by 48-55% on established testing datasets
Why It Matters
Automates one of software development's most manual processes, potentially cutting testing time while improving coverage.