Uses python Hearst pattern code from https://github.com/mmichelsonIF/hearst_patterns_python

USAGE: python3 hearst.py [] [-o ] [-s]

If no input file is given, it will process text read from stdin. If no output file is given, it will write to stdout. The -s flag limits the parrersn to the oritinal (simple) ones from thew 1992 paper.

Output has tab-separated lines like 'specific\tgeneral\tsource_file', e.g., 'apple\tfruit\trecipies.txt'

From python, call hearst.hearstify(, []) to get an iterable of pairs

Works in both Python3 and 2.X

See: Marti A. Hearst. 1992. Automatic acquisition of hyponyms from large text corpora. In Proceedings of the 14th conference on Computational linguistics - Volume 2 (COLING '92), Vol. 2. Association for Computational Linguistics, Stroudsburg, PA, USA, 539-545. DOI: https://doi.org/10.3115/992133.992154 http://people.ischool.berkeley.edu/~hearst/papers/coling92.pdf

apt_2017 is a directory with some sample text files that you can use for testing or experimentation

apt_sg.txt is the output produced by hearst.py run on more than 400 APT documents where the 1st column is the specific term and the second is the generic one.

apt_sg_uniq.txt are those results after processing with sort and uniq

apt_gs.txt is the output produced by hearst.py run on more than 400 APT documents where the 1st column is the generic term and the second is the specific one.

apt_gs_uniq.txt are those results after processing with sort and uniq