27.03.18

Expert Opinion

Meet BERT!

     

    MEET BERT our Beer Encyclopaedia Recommendation Tool

    During Gartner’s evening networking party we were excited to unveil BERT the Bartender! Choosing from a wide selection of beers, BERT used machine learning to recommend the perfect beer to add to      your collection of personal favourites.

    BERT utilises a range of sensory descriptors, such as flavour, colour, and taste, and a similarity algorithm to identify fantastic brews based on the profile of your current favourite.

    This intuitive program is packaged into a Squirro application, allowing us to utilise Squirro’s intelligent search functionality, highly scalable data storage engine, and modern visualisation capabilities. This has allowed us to combine our own Python based data science application with Squirro’s cutting edge context intelligence technology.

    The recommender works by first looking up your beer of choice in our sensory encyclopaedia. It then uses a fingerprint of terms most frequently used to describe that beer (e.g. Citrus, Golden, Strong, Creamy) to determine which other beers have the most similar attributes. So if you tell us your favourite beer, which is usually described to be full-bodied and hoppy, expect BERT to recommend you beers with mouthfeel aplenty or double-hopped delights.

    We hope BERT helps you discover some new favourites and is a fun spin on the bar service that you are usually accustomed to!