To Design and Implement a Recommender System based on Brainwave: Applying Empirical Model Decomposition (EMD) and Neural Networks

Abstract

Recommender systems collect and analyze users' preferences to help users overcome information overload and make their decisions. In this research, we develop an online book recommender system based on users' brainwave information. We collect users' brainwave data by utilizing electroencephalography (EEG) device and apply empirical mode decomposition (EMD) to decompose the brainwave signals into intrinsic mode functions (IMFs). We propose a back-propagation neural networks (BPNN) model to portrait the user's brainwave preference correlations based on IMFs of brainwave signals, thereby designing and developing the book recommender system. The experimental results show that the recommender system combined with the brainwave analysis can improve accuracy significantly. This research has highlighted a future direction for research and development on humancomputer interaction (HCI) design and recommender system.