The Gamium Recommendation Engine is based off one simple concept: you get better game recommendations from people who like the same games as you.
Most game recommendation engines look at the games you’ve rated and try to find similar games to those you’ve rated high. It’s a good first approximation: if I like Call of Battlefield I’ll probably enjoy Shooter: Origins.
And if gamers mostly played the same genre of games over and over again this would be all you needed.
The problem is, game recommendation systems like this “overfit” the data: if you play a lot of shooters, they’ll just recommend more shooters, and you’ll lose out on a whole variety of games.
In reality gamers play games across all genres and just because I enjoy one game in a genre or series doesn’t mean I’ll like others. Gamium avoids this by building a circle of similar gamers and using their reviews to recommend games.
How to recommend The Gamium Way
Step 1: Collect your game reviews.
We need a database of your game review data. We use your game review profile on RAWG.io.
The more game reviews you have, the more accurate our recommendations are.
Step 2: Collect a LOT more game reviews.
We need a large database of game review data from a wide variety of gamers with a wide variety of tastes in games.
This is provided to us via the power of the RAWG.io API. They have millions of reviews by thousands of gamers on tens of thousands of video games.
Step 3: Find similar Gamers Just Like You.
We’re not looking for exact matches, we’re looking for gamers who’s tastes in video games are really similar to yours. Maybe you disagree on the relative merits of a specific game or on the artistic direction of a sequel, but after crunching millions of reviews we can find gamers with similar tastes.
This makes up your personal circle of gamers.
Step 4: Process all the reviews and make predictions.
We process the hundreds of reviews of video games by your personal circle of similar gamers in order to predict what you’ll rate a game as.
We collect both positive and negative reviews and try to find games we think you’ll both like and dislike.
Step 5: Use those predictions to build up a game list.
We provide both recommended games and “disrecommended” games from our database. We’re pretty sure you’ll like the recommended games, and pretty sure you won’t like the others.
Getting Better Recommendations
Our recommendations are only as good as the data they’re based off of. In order to find the best circle of gamers that we can for you we need lots of your game reviews.
The more reviews we have, the better we can make our recommendations.
You don’t need to write reviews, you just need to give them one of the review scores on RAWG: Exceptional, Recommended, Meh, and Skip.
Use the words
RAWG’s game review system uses the words “Exceptional”, “Recommended”, “Meh”, and “Skip” for a reason. They have specific meanings. Exceptional means that a game is the best of its genre, while Skip means you can safely avoid this game. Using these words as a guide, we are more likely to find a lot of game reviewers to compare you with, building you the best circle we can.
With that said, review the games the way you’d want them to be reviewed.
The magic of our recommendation system is we also tend to find other gamers who recommend games the same way you do. If you use “Recommended” to mean “4 out of 5 stars” we’re likely to find other gamers who rate games the way you do. You just limit the number of gamers we’ll be able to find in your circle. But you do you! This is all about personal taste.