ketchum | Use word vectors to interactively generate lists | Natural Language Processing library

This is code for generating lists of similar words, using word vector similarities from the fasttext data, released by Facebook. It's useful for building collections of words to slot into generative grammars, such as Kate Compton's tracery.
The code has a few code dependencies, listed in requirements.txt. You'll need recent-ish versions of the annoy, flask and numpy libraries. pip3 install -r requirements.txt will grab them, if you don't already have them installed. I've only tested it on Python 3, but it shouldn't be difficult to translate it to Python 2 if you're really fussed. Start the server by running python3 ketchum.py. The first time it runs, it will take a while (about half an hour on my machine) to download data and build some indices. You'll need around 2.5GB of free disk space, and a decent amount of RAM. Once this process has run once, subsequent runs should be pretty zippy. The server runs on port 8765 (you can change this at the top of ketchum.py). Point your web browser at http://127.0.0.1:8765/ and have a play around. I'm using the fasttext word vectors, provided by Facebook. There's a lot of weird stuff in there, so watch out for misspellings, non-English words, proper nouns, slang, weird punctuation, and just about any other kind of oddness you can think of. On the other hand, it's extremely comprehensive and should be able to handle just about anything you feel like throwing at it.