EEG-based emotion recognition in neuromarketing using fuzzy linguistic summarization

Abstract

In recent years, to increase market share, companies have preferred neuromarketing over traditional methods for better analysis of consumer behavior. Since it easily detects customers' subconscious preferences, electroencephalography (EEG), a brain imaging method, has become widespread within neuromarketing techniques. To make sense of EEG signals, dimensional models are used to convert them into emotions. These steps can reveal emotions and preferences easily but still require an expert for detailed stimulus analysis. This article proposed a fuzzy linguistic summarization approach to provide a decision support tool aimed at presenting detailed analysis to neuromarketing experts. EEG signals were recorded to analyze a hotel's three (audio, video, web page) advertisements (ads). These were converted into fuzzy emotion labels in a modified Russell's circumplex model for more specific analysis. Then, these emotion labels were used in linguistic summarization. EEG data were handled in three types: univariate, multivariate, and multigranular detected time series. Each ad was summarized according to demographic features, such as gender and age, allowing comparisons between ads and their segments. The granular trend detection algorithm was modified to detect the simultaneous effects of ads. This study will inspire future studies with three innovations: fuzzy linguistic summarization technique in neuromarketing, fuzzy emotion recognition, and a modified multigranular trend detection algorithm that detects simultaneous agglomeration that is often overlooked.