Automatically finding Codenames clues with GloVe vectors

I also tried writing an AI for Codenames a couple of years ago, with the help from some friends. Interestingly, we also found out the hard way that the word &quot;round&quot; aren&#x27;t used in the same contexts as things that actually are round, so our AI also struggled to connect &quot;round&quot; with words like &quot;Jupiter&quot; or even &quot;Sphere&quot;. I&#x27;m not sure how to fix that, but maybe knowledge graphs could help?<p>But I also thought it was surprising how well it worked most of the time! It wasn&#x27;t quite as good as a human spymaster, but with the right choice of scoring function we could actually get a model based on word embeddings to play a pretty reasonable game of Codenames.<p><a href="https:&#x2F;&#x2F;github.com&#x2F;Gullesnuffs&#x2F;Codenames" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;Gullesnuffs&#x2F;Codenames</a>

This is a cool article, and I appreciate the actual testing.<p>I also heartily recommend this implementation using conceptnet embeddings.<p><a href="https:&#x2F;&#x2F;github.com&#x2F;commonsense&#x2F;codenames" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;commonsense&#x2F;codenames</a><p>Grab the latest mini.h5 and go. <a href="https:&#x2F;&#x2F;github.com&#x2F;commonsense&#x2F;conceptnet-numberbatch&#x2F;blob&#x2F;master&#x2F;README.md" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;commonsense&#x2F;conceptnet-numberbatch&#x2F;blob&#x2F;m...</a>

This is cool. With a 300-dimensional word representation, there are a lot of degrees of freedom to play with to achieve sharing meaning axes with the good words and diverging along meaning axes with the bad words.<p><i>There&#x27;s an over-indexing problem: words that happen to be very close to one or two of the targets will rank highly even when they&#x27;re far away from the third. Minimizing the maximum distance from any target helps mitigate but doesn&#x27;t entirely solve this problem.</i><p>This is because they combine the distances using something like more like OR-gate (are any of these very good) and should be using AND (are all of these very good).

Interesting, but in the first example I would be worried that Wok would be misheard as &quot;walk&quot;, which might entice people to pick the instant fail card of &quot;fall&quot;. That&#x27;s a meta-level of the game (which I&#x27;ve played with friends and family) that you have to be aware of, and sometimes you can use to your advantage. Because of that, using a program that can perfectly disambiguate between the word and associations will probably lead to some non-intuitive results sometimes.