Fraud detection of black pepper using metal oxide semiconductor gas sensors

Abstract

Sarawak black pepper, having superior quality deserves attracting a higher market price compared to its competitors. This highlights the importance of establishing product differentiation to protect the economic interest of the farmers of Sarawak black pepper. Therefore, a simple and quick method to authenticate Sarawak black pepper from other countries of origin is required, especially from a fraudulent packaging that falsely claim the origin of others as from Sarawak. In this study, a metal oxide semiconductor (MOS) gas sensors array consisting of four sensors is used to capture the volatile organic compounds (VOC) profile of black pepper berries. The intention is to use this signal as differentiator for the origin determination. A total of 600 black pepper berries were sampled (300 from Malaysia, 300 from India) to generate reference data for fraud detection. Each set of sample data consists of four sensors readings and pepper berry mass, which would undergo dimension reduction using principal component analysis (PCA), followed by Naïve Bayes classification to produce a color map of posterior possibilities for sample prediction in a fraudulent package of black pepper. By testing the prediction model against five mixtures of black pepper, each in different composition of Malaysia (Sarawak) and India black pepper, an overall accuracy of 84.8% is achieved. The results confirms that a MOS gas sensing system integrated with PCA and Naïve Bayes classification analysis methods has a great potential to produce reliable geo-tracing results even in a fraudulent black pepper packaging, hence the developed methodology is suitable for black pepper fraud detection applications.