Stasis - Python wrapper for Semantic Similarity datasets
Under the auspice of the EXPERT project (http://expert-itn.eu/), we have written a python wrapper to the STS datasets and we hope that it helps anyone with easy manipulation the datasets.
If you just need a tab-separated file, you can easily find the
sts.csv available in the same repository. The repo also contains other (maybe) useful datasets that are manually compiled by the maintainer when they are free.
Disclaimer: The repository comes as it is. It should NOT be considered as the official SemEval's (Semantic Textual Similarity) STS data and it is not affiliated with the STS organizers. We've created this so that people can easily do something like
graphlab.SFrame('sts.csv') and work with the dataframes with little hassle.
Below is a list of datasets/wrappers you can find here
- STS: SemEval Semantic Textual Similarity (STS2012 - 2015)
- CLSS: SemEval Cross-level Semantic Similarity (CLSS)
- SICK: Sentences Involving Compositional Knowledge
Please feel free to add datasets/wrappers to the repository. Or post an issue to request for wrappers to the repository.
Please cite the respective references for the datasets when using them in your publication!
If you want to cite this repository, you can cite this paper where we created used the
sts.csv in SemEval-2015