General overview of Streamplate’s real-time recommendation system
What sets apart Streamplate from many other food-based listing services is how it generates recommendations on factors besides just geographical proximity.
Distance to a venue is naturally important when considering where to eat or drink. However, Streamplate believes this is just one aspect of a much richer decision-plane. For Streamplate there’s a set of over a hundred variables that determine the listings of venues and their associated menus.
Some of these factors include the people you’re with, humidity, UV-strength, the flow of traffic, approaching holidays, the instantaneous popularity of a venue, surrounding major events and so on.
When a user opens Streamplate they receive a sorted listing of venues in general order of preference for that user. Within each venue is then a sorted menu, again, sorted by preference. That means that while there’s a serial delineation in logic (first —sort the venues, second — sort the menus), there has to be near-parallel execution.
This is naturally problematic when one informs the other.
On top of this, there’s an importance focus on speed — fast listings produce better user experiences. So it’s not just a matter of relevance, but also timely-relevance.
The above graph displays a general overview of how Streamplate operates. It’s our initial goal to display results that lead to high-conversion rates in less than 2 seconds.
To further emphasise the technical goal — imagine if Google sorted their results for you based on who you were with, where you were and what you wanted in that moment. Then multiply that by the number of users using Streamplate in that moment.
Be sure to sign up for our Limited Release in the coming weeks here.
Thanks for reading!