Negative News Recognition During Social Media News Consumption Using EEG

In this study, we introduce a novel system to early recognize the negative emotional state, which may result from consuming negative news across social media using EEG signals. The system acquired eight symmetrical temporal channels using a custom-design EEG headset. Seven types of features based on Fourier and wavelet transforms were investigated. Support Vector Machine (SVM) and Multilayer Perceptron (MLP) were applied as the classification algorithms. It turned out that the combination of Power Spectral Density (PSD)-based frequency point feature and MLP classifier yielded the highest detection accuracy. The experimental results for ten subjects highlighted that on average, the proposed system could detect a negative emotion as short as 0.5 s with high accuracy of 94%. Hence, our proposed system demonstrated the feasibility and practicality of implementing a real-life application of recognizing negative emotional state during social media news consumption which may help to prevent the users from mental illness suffer.