What is this?

This is a small toolbox to evaluate semantic relatedness scores on standard datasets. Currently, the supported methods are

  • Pearson correlation
  • Spearman correlation
  • Cosine measure

each with an option to add weights for the data. The repository also includes standard datasets, as described on my blog.

What do I need?

Requirements are

  • Python 3
  • NumPy
  • Pandas

How do I use it?

Just execute

python install

in the src/ directory to add the project as an egg to your Python distribution.

Are there bugs?

Yes, there are. Feel free to report them at the issue tracker. I will be working on this project from time to time, so you might want to stay up to date.

No Documentation?!

Maybe I will add some, maybe I won't. Should be mostly self explaining.