Project abstract

One of the most important questions in enology and prevention of wines adulteration are related to the correct assessment of geographical origin, cultivar and vintage. Apart from the continuous improvement of analytical methods, an important aspect is related to the development of reliable data processing tools. Metabolomics represents the comprehensive and nonselective analytical chemistry approach aimed to prove a global description of all the metabolites present in a biological sample at any given time. The main methods of metabolomic analysis rely on spectroscopic detection. Two basic approaches are used in metabolomics: target and nontargeted (profiling and fingerprinting) analysis. Presently, the main analytical techniques that are used for food metabolomics studies are nuclear magnetic resonance (NMR) as well as mass spectrometry (MS)- based techniques. Beside these, Raman spectroscopy also proved a high potential for the development of new recognition models for food and beverage differentiation and starts to benefit by an increase interest from research and control laboratories in this regard. Nowadays, an emerging step in the development of food and beverages authentication tools is given by the application of artificial intelligence (AI) for the construction of recognition models with respect to predefined criteria (i.e. geographical origin, cultivar, vintage). The project aims the development of an innovative web application for wine authenticity control which will rely on new recognition models for wine discrimination according to geographical origin, cultivar and vintage. The models will be built using artificial intelligence (AI) for the advanced data processing of the ¹H-NMR and Raman spectra. Different Machine Learning algorithms (ML) will be used for the models’ development. The approach based on which the best recognition models will be built will be further used for the web application development.