COS960 is a Chinese word similarity dataset of 960 word pairs. Each pair of words is annotated by 15 native speakers with a similarity score which reflects true similarity. The 960 word pairs are further divided into 3 groups according to their Part Of Speech tags, including 480 pairs of nouns, 240 pairs of verbs and 240 pairs of adjectives.

To use COS960 to test your word embedding, use command

python correlation_calcu.py {VECTOR_FILE}

The data in the files is formulated as

[Word1] [Word2] [Average] [Annotator1] ... [Annotator15]

小心谨慎  谨慎小心     4.0         4      ...       4 

If you use the dataset, please cite this:

@article{huang2019COS960,
Author = {Junjie Huang and Fanchao Qi and Chenghao Yang and Zhiyuan Liu and Maosong Sun},
Title = {{COS960: A Chinese Word Similarity Dataset of 960 Word Pairs}},
journal={arXiv preprint arXiv:1906.00247},
Year = {2019},
}