Vi Case Study: Food Ontology Data Set Creation

Intro

Drawing upon its own extensive database, compiled by expert sommeliers and food researchers, the Vi app uses machine learning to suggest the perfect wine accompaniment for any dish. Aimed at individuals and businesses, the app demystifies the wine selection process.  

Challenge

The initial challenge for Vi was the creation of a dedicated food/wine database – especially when accounting for variances in names and ingredients across dishes. For the final product to be suitably tailored to each customer, the database required careful allocation of sub-categories, with wines split between regions and foods by either cuisines or ingredients.

Action Steps

Using a top-down approach, the research team compiled a dish database emphasizing foods catering to American audiences and the appropriate wine pairings.

We refined the database using specific categories: cuisines, key ingredients, dish types, and flavor profiles.

Finally, we marked synonyms to enhance the search functionality. Creating a comprehensive and easy-to-navigate food ontology database, allowing users to map each wine pairing to specific dishes, regions, ingredients, and cuisines.

Results

Following the successful implementation of our food/wine ontology, Vi launched its first web application. This application allows customers to pair food and wine using data produced by food experts and sommeliers.

Vi can now demonstrate specific pairings to customers according to various categories, including wine type, regions, grapes, dishes, ingredients, and budget.

Learn more about how we can help you better collect and organize data for your team by scheduling a quick call.