A compatibility test method for Android applications based on multimodal large models in crowdsourcing scenarios

With the popularity of smartphones, the Android operating system has taken a significant share of the global market, prompting researchers to explore automated testing techniques to cope with the diversity and complexity of operating systems. This paper proposes a compatibility testing method for Android applications based on multi-modal large model, which combines Android automated testing tools and multi-modal large models to achieve simultaneous testing of multiple mobile phones. The multi-modal large model named STLLM proposed in this paper can automatically generate software test scripts based on Android system, support simultaneous testing on multiple devices, solve the compatibility problems and problems caused by manual coding in software testing, and improve the efficiency and accuracy of testing. Experimental results show that the proposed method can effectively improve test coverage, reduce test cost, and save a lot of time and energy for developers.