How do we measure what will make you happy?

eezy’s main intention is to maximise the happiness of every user, every day. Reducing choice is just one piece in the puzzle of happiness. Puzzle of happiness…yep, that’s what I just wrote. This article will elaborate on how important it has been for us to know exactly what things are in order to recommend them.

What are candidates and features?

We refer to any option we could recommend that we think will be enjoyed — a movie, restaurant, concert or even just the idea of meeting a friend — as a candidate. We really do think of them as candidates to make our users happy. Every candidate has attributes, or as we call them, features. Features have to be mathematical representations of quantitative or qualitative data points. Let’s take a movie. Quantitative features would be things like the genre, location or starting time of the movie. Qualitative would include features such as the way people speak about a movie in online reviews or the personality types of those reviewers.

So the more features a candidate has, the better they are described, and therefore the more the AI is able to compare them. All features have probabilities assigned to them, to measure how sure we are about it having this feature (as it’s essential that the features of a candidate accurately reflect it). The candidate/feature pair is called the “candidate space”.

What do we look at?

The features that make up a candidate don’t come from one place, they are found all over the internet. We have built an aggregator that finds all these data points and creates bespoke features out of them. Let’s take a restaurant, we analyse reviews, the tonality of the person writing the review, the rating they gave and the exact things they spoke negatively about. In addition to this we look at menu items and their prices, location, how busy a place is and countless other factors. From this, we create the candidate space — a multidimensional space comprised of several hundred dimensions that describes this candidate. Everyone at eezy believes that the human brain should not be burdened with several hundred dimensions…

These tasks are designed for and should be executed by machines. Experiencing a Tarantino movie or the new fusion sushi, on the other hand, is for us humans to enjoy executing.

Features continuously change while our data aggregator feeds new information into them. This means that we can further personalise our recommendations over time. We use all the features of our candidates to learn what is important to our individual users. The more the user uses eezy, the more we know and therefore the better our recommendations become.

The human mind is made for intellectual creativity, not for optimisation problems. This is why we created eezy, to make your life… well, eezy.

It has always been our founder’s core belief that in order to be happy you need to expand your horizons and explore new things. The psychologists agree. So we have designed our algorithms to always search for new things in candidates that users may like, in order to lead users into new, exciting experiences. In AI this is often known as exploration vs. exploitation — a topic we are extremely passionate about.

Whilst we are at the beginning of our journey, we believe that our product speaks for itself. Download it here

— — — — — — — — — — — — — — — — — — — — — — — — — — — — — — —

Watch this space as we take you through our startup journey and share some more insights.