https://github.com/stitchfix/context2vec

Using Word2Vec on lists and sets
https://github.com/stitchfix/context2vec

Last synced: about 1 year ago
JSON representation

Using Word2Vec on lists and sets

• Host: GitHub
• URL: https://github.com/stitchfix/context2vec
• Owner: stitchfix
• License: mit
• Created: 2015-05-16T00:51:23.000Z (about 11 years ago)
• Default Branch: master
• Last Pushed: 2015-10-28T22:34:36.000Z (over 10 years ago)
• Last Synced: 2025-03-30T13:03:46.605Z (over 1 year ago)
• Language: Python
• Size: 137 KB
• Stars: 34
• Watchers: 165
• Forks: 6
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE
  • Code of conduct: CODE_OF_CONDUCT.md

Awesome Lists containing this project

README

          
## Using Context2Vec

Use Word2Vec on lists and sets

Use like:

    from sphere.models.context2vec import ContextCorpus

    from gensim.models import Word2Vec

    # If user 1 bought items 0, 1, 2, 3 and user 2

    # bought 3, 4, 5 and user bought item 10

    contexts = {1:{0, 1, 2, 3}, 2:{2, 3, 4, 5}, 3:{10}}

    cc = ContextCorpus(contexts)

    # Run the w2v model

    model = Word2Vec(cc, size=30, iter=100, min_count=1, alpha=0.025, sg=0)

    # Find the most similar user to user 1

    results = model.most_similar('C1', topn=100)

    related_users = filter(lambda x: x[0].startswith('C'), results)

    # Find all user vectors

    users = filter(lambda x: x.startswith('C'), model.index2word)

    user_vectors = [model.syn0[w2v.vocab[w].index] for w in users]

## Running tests

Simply run

    nosetests

Or to debug:

    nosetests --pdb