Code and data for "Learning Scalar Adjective Intensity from Paraphrases"

If you use the code or data in this repo, please cite the following paper:

@InProceedings{D18-1202,
  author = 	"Cocos, Anne
		and Wharton, Skyler
		and Pavlick, Ellie
		and Apidianaki, Marianna
		and Callison-Burch, Chris",
  title = 	"Learning Scalar Adjective Intensity from Paraphrases",
  booktitle = 	"Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing",
  year = 	"2018",
  publisher = 	"Association for Computational Linguistics",
  pages = 	"1752--1762",
  location = 	"Brussels, Belgium",
  url = 	"http://aclweb.org/anthology/D18-1202"
}

Directory contents

• The adjective graph that encodes adjectival paraphrases from PPDB of the form (RB JJ1 <--> JJ2) is under jjgraph/
• Code for computing a pairwise paraphrase-based intensity score is in pairpredict/
• Code for the downstream tasks of global ordering and indirect question answering are in globalorder/ and iqap/ respectively
• The crowdsourced dataset of adjective scales is in globalorder/data/crowd/gold_rankings. Each file contains adjectives belonging to one scale, where they are ranked from least to most intense by line.