****************************************
*  Adjective-Noun Composition Dataset  *
****************************************

Contained is the dataset used for the analysis and system evaluations in our submitted paper:
"Most "babies" are "little" and most "problems" are "huge": Compositional Entailment in Adjective-Nouns." 

This data is licensed under a Creative Commons Attribution 3.0 United States License. If you use this data in your work, please cite the following paper:

@article{PavlickAndCallisonBurch-2016:ACL:Adjectives,
  author =  {Ellie Pavlick and Chris Callison-Burch},
  title =   {Most baies are little and most problems are huge: Compositional Entailment in Adjective Nouns},
  booktitle = {Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (ACL 2016)},
  month     = {August},
  year      = {2016},
  address   = {Berlin, Germany},
  publisher = {Association for Computational Linguistics},
}

************
* CONTENTS *
************

This archive contains the following files:

AN-composition-dataset/
  README -- This README 
  addone-entailment/
    all-data.txt -- Our full dataset of AddOne RTE sentences annotated on MTurk
    splits/
      data.train -- The training split used for our results in Section 7 of the paper
      data.dev -- The dev split used for our results in Section 7 of the paper
      data.test -- The test split used for our results in Section 7 of the paper
  natural-logic-relations
    labeled-data.txt -- ANs in context labeled with 5 natural logic relations, used for analyses in Section 6 of the paper.

All of the add-one files are in the following (tab-separated) format:
mean_human_score  sentence_id  adjective  noun  premise  hypothesis

The natural-logic-relations file is in the following (tab-separated) format:
genre  nl_relation  mean_human_score_insertion  mean_human_score_deletion  entailment_insertion  entailment_deletion  adjective  noun  context


***********
* CONTACT *
***********

If you have any questions, comments, complaints, or deep philosophical thoughts about the nature of 
adjective noun composition, I would love to talk to you! You can contact me at elliepavlick@gmail.com. 