API Parameter Recommendations Made Effective
Wenhua Yang · 2020
The determination of actual parameters is indispensable in applying API methods that require parameters. While researches on the problem of API parameter recommendation are emerging, they all rely on simple parameter usage patterns. In this paper, we propose a program analysis and machine learning combined method for recommending API parameters. This approach first uses the slicing technique to capture complex usage patterns of the parameters and then employs deep neural networks to recommend API parameters based on the slices.