|
|
French researchers from the University of Lyon have developed a method that enables them to identify the origin and grape varieties of wines. Based on the analysis of mineral profiles of wines and using an AI tool, the procedure can assign the countries of origin of the wines with an accuracy of 92 percent. The method is also very successful in recognizing French wine regions (91%) and the main grape varieties (85%).
The procedure utilizes the mineral components of a wine, referred to as the "mineral wine profile" (MWP). This remains stable regardless of aging or storage conditions. A total of 41 chemical elements are analyzed, which allows for a more accurate capture of a wine's terroir. Since minerals from the soil enter the grapes and are less influenced by the production process, the MWP forms a reliable and consistent source of origin information. With the help of a machine learning model, the team was able to build a comprehensive database of 12,966 international wine samples. The AI tool was trained to recognize wines based on a 30 ml sample with a reliability of over 99 percent.
According to the researchers, this method represents an efficient and scalable alternative to previous techniques such as DNA and organic profile analyses, which are often affected by aging and storage. Mineral profiling could play an important role in the fight against wine fraud and in ensuring the authenticity of wines on a global scale.
(ru / Oene One - Photo: Raffaella Usai)