A Feature and Deep Learning Model Recommendation System for Mobile Application
Xiangying Che, Qibo Sun · 2021
Deep learning is becoming more and more popular in mobile applications. We also find apps using deep learning are more popular. For app designers and developers, it is particularly important to filter out certain features of corresponding scenarios and target user groups effectively from complex and diverse deep learning techniques, which is hard for developers. To solve this problem, we design a mobile deep learning application features and models recommendation system for app designers in this paper. Designers only need to input the description information of their app to acquire some appropriate features and models advice. Firstly, we analyze the feasibility and necessity of mobile application feature and model recommendation. Then the database of app details is built and sorted out to provide the basis for the recommendation system. Finally, we give out a new recommendation system specially designed for mobile apps on Google App Store, which mainly includes Labeled-LDA text classification and BERT text similarity matching combined with recommendation correction. The demonstration shows that our system has some reference value for mobile application developers.