Autoencoder based API Recommendation System for Android Programming

As a typical example of modern Information Technologies, Android platform and Apps are widely used by smartphone users all over the world. Thus, the research of designing models for assisting programmers in writing Android codes is of great importance and value, and recommending API usages is a stereotype task in this aspect. This paper applies Autoencoder neural networks into the model of API recommendation system for Android programming, and designs new Autoencoder based Android API recommendation system. This paper carries out experiments on the collected Android code dataset and verifies the effectiveness of the newly designed models compared with classical recommendation models.