Hypernym-LIBre: A free Web-based Corpus for Hypernym Detection [PoS and Dep-parsed Version]
Shaurya Rawat, Mariano Rico, Óscar Corcho · Zenodo (CERN European Organization for Nuclear Research) · 2020
Hypernym-LIBre ( DOI: 10.5281/zenodo.3662204 ) is a free Web-based corpus for Hypernym detection. Here we provide its part-of-speech tagged and dependency-parsed version. Both have been annotated in the text as follows: For every word in the corpus, there is a PoS tag and dependency annotation separated by a '-'. Example: corrected-VERB-ROOT where corrected is the raw word followed by its PoS and dependency annotation. The file is provided as a zip file: Compressed File Size: ~15GB Total Size: 80.6GB uncompressed Number of Files: 442 files, ~180MB each 10.5281/zenodo.3695237 hyponym-hypernym pairs extracted from Hypernym-LIBre using Hearst patterns