Distributional Semantics

This is a WIP Python 3 library of functions for learning vector space representations of words.

Functions

As of now, this library offers functions for:

Processing corpora

• Feature selection
• Co-occurrence matrix (with or without decreasing weighting)

Learning vector space representations of words

• PPMI+SVD
• COALS
• GloVe (with regular stochastic gradient descent instead of AdaGrad)

Measuring semantic similarity between words

• Cosine
• Generalized Jaccard

Evaluating semantic models on a word similarity task

• Ground truth
  • WordSim-353
  • MC
  • RG
• Correlation
  • Pearson
  • Spearman

Requirements

NumPy
SciPy
scikit-learn