A smartphone-based activity-aware system for music streaming recommendation

Contextual information is helpful in building systems that can meet users’ needs more efficiently and practically. Human activity provides a special kind of contextual information that can be combined with the perceived environmental data to determine appropriate service actions. In this study, we develop a smartphone-based mobile system that includes two core modules for recognizing human activities and then making music streaming recommendation accordingly. Machine learning methods with feature selection techniques are used to perform activity recognition from smartphone signals, and collaborative filtering methods are adopted for music recommendation. A series of experiments are conducted to evaluate the performance of our activity-aware framework. Moreover, we implement a mobile music streaming recommendation system on a smartphone-cloud platform to demonstrate that the proposed approach is practical and applicable to real-world applications.