36

What exactly is the mechanism behind this statement?

screenshot of the Steam interface showing a part of the Steam Store's page for Cyberpunk 2077

The game in question, Cyberpunk 2077, at the moment of writing has a general rating of 76% ('Mostly Positive') — I would not consider that 'love'. But if "players like me" love this game, then a specific group of Steam users must have a much higher rating for this game.

Who are these "players like me"? What parameters determine which users fall among my gaming kindred?

Alternatively, how does Steam define 'love'?

6
  • What part of the Store are you seeing this in? Does it show up on your Store Home and say "Recommended because you played games tagged with"?
    – Malady
    Apr 20 at 15:45
  • No, it can be seen on Store pages of games that happen to be loved by players like you, so 'results may vary'.
    – Joachim
    Apr 20 at 15:46
  • 1
    @Malady here is another example of where it is
    – Timmy Jim
    Apr 20 at 15:50
  • 6
    'a specific group of Steam users must have a much higher rating for this game' -> I want to remind you that a steam review is either "recommand" or "do not recommand". So a rating of 76% doesn't mean that people "kinda liked it", it means that 76% of people that played it would recommand it. So there's no "rating" in how much people liked it individually.
    – Echox
    Apr 21 at 7:30
  • It just sounds like a recommender system, like you'd find on Facebook, TikTok, YouTube and many other places. Individual recommender systems may work very differently, but these details typically aren't exposed to the general public, or it's exposed to some degree through a technical description that probably wouldn't mean much to the layman.
    – NotThatGuy
    Apr 21 at 7:59

2 Answers 2

40

A few parameters are made clear through Steam's Interactive Recommender, which can be found in the Steam Labs:

Screenshot of Steam's Interactive Recommender, highlighting the logic used by the recommendation engine: gameplay history, including playtime, a slider for how much to weight recommendations by popularity, and another slider to filter by the age of the game.

Its description mentions how:

This experiment looks at how much you've played each game in your Steam library, and uses the magic of machine learning to recommend games you might like. Filter your results by picking games that are popular or niche, and drill down by release date and tags.

So it seems to consider playtime as a large factor (as it logically determines one's favourite games), and magic! ...of machine learning.

One the Steam page introducing this then new feature, it reads (my emphasis):

How It Works

The Interactive Recommender uses a machine learning model that is trained based on the playtime histories of millions of Steam users. It's not directly affected by tags or reviews — it instead learns about the games on Steam by looking at what users actually play. The basic idea is that if there are other players with similar play habits to you, who also play a game that you haven't tried yet, then that game is likely to be one you'll enjoy too.

We're also starting to apply the underlying model in other parts of the Steam store, where we think it can help players see the most relevant content or make more informed choices. For example, when viewing the page for a particular game, you may sometimes see "Players like you love this game" shown as a reason why the game is relevant to you, alongside other factors.

Not particularly clear, but a good start.

5
  • 1
    Maybe move the Steam Labs quote up, so my new link makes more sense, and the "magic of machine learning" quote source is clearer?
    – Malady
    Apr 20 at 15:56
  • Warning: Steam never forgets the games that it uses to recommend you other games. You can go to any extremely level to wipe it from your history, but it will always remember your otherwise-forgotten history when trying to recommend you games.
    – J. Mini
    Apr 21 at 20:25
  • Since you're not accepting your own answer, what other information are you wanting for an acceptable answer?
    – Malady
    Apr 23 at 18:12
  • @Malady I think for now I wanted to leave it open to see if others are willing to find out more about it. I'd like to heave more insight into the process of the machine learning: what information is being cross-referenced, what is decisive for 'loving' a game (playtime, but also picking a game up again after a certain amount of time has passed, for example?). Things we might never know :)
    – Joachim
    Apr 23 at 21:47
  • 1
    Well, if you have any free games on your account, you could spawn a new account, just install that single game, keep it open, and browse the Steam Store for games that both your accounts Might Love. ... I wonder if Free Soundtracks like Unhack's count for Might Love, not just games...
    – Malady
    Apr 23 at 22:09
9

Well, looking around the 'net, it seems that that bit of the recommendation system is the Interactive Recommender:

The Interactive Recommender uses a machine learning model that is trained based on the playtime histories of millions of Steam users. It's not directly affected by tags or reviews—it instead learns about the games on Steam by looking at what users actually play. The basic idea is that if there are other players with similar play habits to you, who also play a game that you haven't tried yet, then that game is likely to be one you'll enjoy too.

We're also starting to apply the underlying model in other parts of the Steam store, where we think it can help players see the most relevant content or make more informed choices. For example, when viewing the page for a particular game, you may sometimes see "Players like you love this game" shown as a reason why the game is relevant to you, alongside other factors.

0

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .