New Test Ensures Tea Transparency

Singapore’s innovative mindset has produced a potential world pace-setter in tea authentication. Teapasar is a startup online tea  marketplace that launched in Sept. 2018, announcing its technology-based set of tools that fit superbly into the growing priorities of transparency along the bush to cup logistics chain. This starts with certification that a tea is exactly what it is claimed to be in terms of origin, type, date of harvesting, freedom from adulteration, and non-counterfeit. Singapore has a long reputation as an innovation hub. It is a tiny city-state that plays a large role in international trade and targets growth industries through close public and private sector collaboration. One of its priorities is the broad area of digital transformation, in banking, healthtech, remanufacturing, foods, and media, to list a few examples. Teapasar describes itself in very Singaporean terms as “a multi-brand tea marketplace” and a “gateway to tea brands from all around the world”—a trade hub. Its Teapasar service is intended to attract vendors and tea lovers to the hub. The two main components are ProfilePrint, which profiles teas, and TasteMap, which profiles a customer’s tea preferences and matches them with the closest tea options. The innovation does not break new ground in terms of technology or science, but selects from proven tools that it packages into a cost effective, practical combination. The website is attractive, though not yet up to standard for e-commerce customer interaction and responsiveness. The range of teas offered is impressive and marked by a variety of intriguing flavors and blends. ProfilePrint

ProfilePrint creates a metabolomic fingerprint of a tea.  It classifies tea’s origin, terroir, cultivar, harvest date and other identifiers and is an emerging science of its own within molecular biogenetics. The basis is to extract molecules and atoms from a food to produce a snapshot of the metabolic state of the organism at a given moment. Think of it more as blood testing than fingerprinting. A sample of the compound to be tested is injected into what is basically an oven where it is ionized to separate the molecules. The atomic fragments are carried by an inert gas stream such as helium and pass through detectors in the mass spectrometer that graph the fragments’ mass-charge ratios. The software draws on multivariate statistical analysis methods that demand substantial computing power. This specialization is termed chemometrics—analogous to psychometrics. Results are displayed in many forms; the teapasar TasteMap graphical representation is just one, which is tailored to tea drinkers not scientists. There are multiple methods and levels of analysis of specimen sampling used in metabolomics. Amplified DNA fragment length polymorphism (AFLP), for example, works with the genetic coding of cells. Advanced techniques can detect levels of the most exotic molecular ingredients, such as p-coumaroysolglucosol-rhamnosylgalactoside. Many of these have been successfully applied in test settings to tea authentication. One study for instance was able to trace differences between raw and cooked puer and their parental leaves. Another was more than 99 percent successful in discriminating among teas of multiple countries of origin and picking out counterfeits and adulterations. These and comparable studies of Japanese cultivars, herbal blends, and Chinese cultivar pedigrees date back a decade or more and demonstrate proof of concept: biological markers and gene profiling are practical and reliable tools in achieving what fingerprinting provides in criminal investigation and security:  unique identification. Teapasar adds proof of application and of cost-effectiveness. Gas chromatography machines suited for simplified and streamlined methods sell for as little as $2 thousand, with more advanced ones costing around $20 thousand. The Teapasar design can incorporate developments and add computing power in any of its components: hardware, software, data and methodologies. This modularity makes it a platform for expansion rather than a fixed system. TasteMap TasteMap is a separate modular component. It profiles the tea drinker via a short online user-submitted selection of taste preferences in eight categories, including sweetness, richness, and astringency of taste and aftertaste. It then estimates the best matches between tea and customer. The key here is to use artificial intelligence (AI) machine learning techniques to improve predictions via trial and error feedback. This is the core of AI application. TasteMap relies on data training. The model is provided with an initial set of categorized examples and then fed random new samples. Successes in matching get reinforced in the software’s future assessments. The data needed can be immense. To give a typical illustration of the demands for exploiting the full capabilities of a machine learning application, the data training for a robot hand that simply picks up and rotates blocks required the equivalent of 100 years of human trial and error learning and two days of nonstop multiprocessor computer time. One of the lab studies of tea authentication used more than one million data training samples. Teapasar has started with a subset of around 400 data points from 350 tea samples. As it adds vendors and customers, the machine learning will become exponentially more effective. Data is the new currency of AI. Teapasar will succeed the more it attracts vendors and customers to enrich and enlarge that currency. Development Teapasar has built a platform that can be expanded to create a business. Its development has three foundations: The National University of Singapore Food Sciences and Technology Program (NUS) was established by the Singapore government development agencies to leverage the growing national food manufacturing, packaging and export industries. It is a hub for many food companies’ researcher centers. NUS applied its expertise in chemical fingerprinting to develop the profiles of around 300 teas from 35 vendors. The accuracy rate in matching a sample to the profile is 98 percent. The choice among the many methods put a priority on the speed, throughput and cost effectiveness. Medical research for tea requires far more complex equipment and biochemical procedures. The Agency for Science, Technology and Research (A*STAR) provided the machine learning algorithms and data training. A*STAR is government-run and provides support for Singapore’s economic growth innovators. It includes 16 biomed and physical sciences research institutes and 6 centers and consortia. Intriguingly, one of its 2018 announcements was a significant success in using green tea EGCG molecules as carrier cells for biogenetic drugs. A*Star provides the small startup tea company with ultra-advanced technology and expertise.

The Singapore tea marketplace: Singapore has long been a foodie paradise. Its restaurants and street vendor food stalls incorporate many Asian styles and piquant flavors. This tradition has been picked up by its rapidly growing tea industry. Teapasar’s product list includes classical Japanese, Chinese and Indian teas but also some of the local variants that include nasi lemak, an aromatic mix of hojicha and genmaicha from Japan, coconut flakes and dried Singapore pandan and chili. It is difficult to assess the likely future of Teapasar. The company is small and new. There may be very substantial financing demands in expanding the initial system to make it an international service attractive to vendors, growers and customers. Marketing and technical support will demand strong managerial capabilities. Customers will decide if the personalization of TasteMap is of premium value. Regardless, this is a true and welcome innovation that may be a pivotal addition to the flood of first-rate work in tea science and biotech, supply chain integration and transparency. It also signals the value of collaborations. Perhaps it will soon be a necessity, among tea growers, packagers and sellers with biotech, software and food science specialists. Innovation hubs provide models and Teapasar an example of the avenues open to small firms. Source: PLOS, Semantics Scholar, Food Navigator