Tip of the Tongue

Interact with words in a meaningful way.

Comment

Inspiration

Ever struggled to think of a word, but found the thesaurus to be useless? It happens to us all the time. But also, we wonder about how:

  • words are related,
  • what images do these words conjure, and
  • what dank definitions does Urban Dictionary have for our words?

What it does

Tip of the Tongue (TotT) takes as input multiple words (separated by spaces, no idioms, sorry!) and uses them to generate a network of word relationships. For example, in a thesaurus, words related to the input are synonyms, so TotT would gather the synonyms (and synonyms of synonyms, and synonyms of synonyms of synonyms), but we also gather visual relationships from Giphy and other relationships from Urban Dictionary. TotT then agglomerates the results based on a normalized Z-score for each word. The result is a set of words related to the input, but in a different way from a conventional thesaurus. We also return gifs, because why not?

How we built it

Links between words are collected from a variety of sources:

Data is processed via Numpy and NLTK.

Django is then used to prepare and deploy HTML, CSS, and Javascript.

All of this is hosted on Amazon Web Services.

Challenges we ran into

Making Django play nice with Javascript, Python, and Amazon Web Services.

Unicode.

Spaces and URLs.

What's next for Tip of the Tongue

More data sources for word relationships. Graph visualization. Definitions alongside words.

+ 3 more
Share this project:

Updates