Semantic Cursor


Semantic Cursor is a small multi-user game I made as a final project for Machine Learning for the Web.

My goal was to build a playful exploration of a sematic space. I was also interested in exploring the use of conversational interfaces for collaboration.

Users accessing the app share a single game of Semantic Cursor. The goal is to traverse the semantic space, word by word to get from one random word to another random word.


  • it is annoying when someone interrupts you
  • turn taking is impossible to signal
  • there are some bad words in there
  • the random problems are more solveable than expected
  • the cognitive load is heavier than I expected
  • there are some weird pockets, eg a very large space of male first names

Future Opportunities

  • Signaling others (presence, activity)
    • active user count
    • other user currently typing signal
  • Physics
  • blacklisting scary words
  • retraining Word2Vec with a topical corpus specific to different communities
  • more game mechanics

Andrew Lee lives and works in Brooklyn building data products

Currently Hashboard
Previously Apple, Flatiron Health
NYU ITP Class of 2020


© 2024 Andrew Lee