Welcome to BetterRelations

BetterRelations is a game collecting association strengths for scientific purposes.

Start the Game

This is a beta version. It works best with Firefox, but feel free to use other browsers.
If you encounter any problem please tell me.

How to Play

  • The game starts by finding a partner for you. You and your partner are a team for 2 minutes.
  • In each round you will see a topic and two items.
    Your task is to decide which item would've come to your partners mind first when thinking about the topic.

    • What would come to your partners mind first?

    Barack Obama
    ?
    id0
    birth place
    Honolulu
    can't
    decide
    id1
    order in office
    President of the United States
    both nonsense
    321
    1:23
  • If your decision and the one of your partner match you are both rewarded with points.
    If you have multiple matches in a row you will get a combo giving you more and more points on each match.
  • A mismatch won't get you additional points and you'll lose a part of your combo.
  • In case you have no idea what the topic is about, you can get a short description by clicking the ?.
    As you didn't know about the topic you can't get points in this round, but you can still save your combo if you have a match with your partner.
  • If you think that both suggested items are irrelevant or you doubt your partner will know any of them, click the can't decide button.
    In case of a match with your partner you will not lose your combo, but you won't get additional point either.
  • It might also happen that both suggested items are nonsense.
    We apologize for these situations, but one purpose of this game is to identify such errors.
    Please click the both nonsense button in this case to flag erronous items. As above a match on both nonsense will not reward you with points while allowing you to keep your combo.

Background

Thanks to the Semantic Web and the Linked Open Data project, we nowadays have a couple billions of machine usable facts in form of (subject, predicate, object) triples.

A small example if you don't know already: There are quite a lot of facts available about Barack Obama (for example DBpedia knows a couple of facts about him).

DBpedia collects these facts (as the name might already have told you) by parsing Wikipedia articles. Nevertheless, two large problems exist:

  • While it's very nice to have so many facts available (for example to answer some complex questions) often you're only interested in the most important ones (for example if you want to explain to someone who Barack Obama is you don't want to confront her/him with 300 randomly ordered facts).
    Currently this is not possible. You can't ask DBpedia to just give you the 10 most important facts.
    Hence, the task of this game is to collect such a "common sense ordering" for a couple of topics in background. Once we gathered a bunch of such ordered lists, we hope to be able to generate and evaluate heuristics, so we don't have to "play" all the billions of facts.
  • Another problem is that the extraction process of the DBpedia is not perfect and introduced many errors.
    We hope to be able to identify systematic errors as a side-effect of this game.