There’s a name for it: tea fraud. A number of high profile scandals in recent years, such as China-made teas with cancer-causing colouring and cheap teas masquerading as expensive brews, have made people cautious about their once-relaxing cuppa.
To protect consumers from health issues caused by drinking these teas, governments are relying on QR codes on packages or sending samples to labs for testing – but it’s a costly and slow process.
Using machine learning, a Singapore-based site hopes to give tea-drinkers worldwide a way to buy teas that are authentic and safe.
Called teapasar, the tea-related e-commerce platform will match vendors’ stock against the taste profiles of hundreds of authenticated teas – including rare or exclusive blends.
Researchers from the National University of Singapore’s Food Science and Technology Programme recorded these profiles for the site using chemical fingerprinting technology. Each tea has a unique taste map, recording factors like sweetness, richness, bitterness and aftertaste, as well as its origin and harvest date.
Professor Zhou Weibiao, Director of the NUS Food Science and Technology Programme, said that the technology is a “fast and reliable” way to authenticate teas and verify labels in an era of fraudulent products.
Teapasar also worked with research agency A*STAR to create software that can predict the taste of a tea based on its organic composition. The researchers will continue to develop the software using established protocol for fraud prevention to make it work faster.
Said the site’s co-founder, Alan Lai: “Teapasar fills a much needed social and ethical gap in an industry that has lasted for centuries… [and] sets a new benchmark for fair practices in tea trading.”