A research team at Kyungpook National University (KNU) has developed a method to determine the soil which can cultivate ginseng consistently in advance by using machine learning.
Professor Jae Ho Shin’s team at Kyungpook National University’s School of Biosciences has developed “a method of determining ginseng crops using soil microbiome and machine learning.”
Even if the same ginseng seeds are planted, ginseng grown in Korea’s soil has superior main ingredients and efficacy than ginseng from overseas countries such as China, so the cultivation soil plays a big role in ginseng quality. Ginseng is a crop that is severely damaged by a series of crops that cannot be used again for more than 10 years once it is grown. However, despite various soil analysis methods, it is considered very difficult to determine in advance whether there will be a series of damage to a particular soil.
Professor Jae Ho Shin’s team obtained more than 100,000 microbial information per sample using next-generation sequencing technology (high-speed sequencing of dielectric material as one of the methods of genetic analysis). It produced a model that identifies 13 million big data as support vector machines (SVM) based on machine learning. In other words, it has developed a machine learning model that can predict the occurrence of ginseng rusty root (GRR) disease before planting ginseng. With this technology, a 90.99% chance of damage can be predicted by microbiological analysis without analyzing past cultivation records of land or soil components.
Professor Jae Ho Shin said, “For ginseng farmers, finding land that has never been planted and renting ginseng is a big problem that influences years of farming. However, it has been almost impossible to prove that ginseng has never been planted scientifically so far, and the conflict is frequent because we can only trust the landowner’s word. “As we observe a person’s microbiome, we can predict the future of the soil with soil microbiome. The technology developed by the research team creates an artificial intelligence algorithm that analyzes soil microorganisms, which can determine whether ginseng has ever been planted with an accuracy of about 91%. The accuracy of the model has room for improvement if it costs more to get more samples.”
The findings, which showed the possibility of artificial intelligence being used in the agricultural sector were published in a cover paper on July 28 of the Journal of Agricultural and Food Chemistry, an international academic journal.