The Card-660 dataset
--------------------

The package contains the following files:

   * README.txt
     this file

   * dataset.tsv
     a tab-separated file, containing all the 660 words pairs and their assigned gold score

   * scores.tsv
     a tab-separated file, containing all the scores of all the 8 annotators
     each column stands for one annotator; the file is line-aligned with dataset.tsv

   * similarity_scale.txt
     the similarity scale used for the annotation of the dataset

--------------------------------------------------------------------

For more information, please see:

Card-660: Cambridge Rare Word Dataset - a Reliable Benchmark for Infrequent Word Representation Models. 
M.T. Pilehvar, D. Kartsaklis, V. Prokhorov, and N. Collier.
EMNLP 2018.

https://pilehvar.github.io/card-660/
