Resources for Coreference Based Embeddings


The following resources can be used to create and evaluate coreference based embeddings.


Coreference chains of English Gigaword data [1]: coreference chains
The chains have been automatically extracted using the Stanford CoreNLP tool [2].

List of words with antonyms: antonyms
This resource is a collection of 6225 English word-antonym pairs from the online dictionary Merriam Webster (

Official train, development and test set for antonym detection: trainDevTestSet
This resource includes positive and negative examples for word-antonym-pairs (denoted by the flags 1 and -1, respectively). It is the same data and split as used in the EMNLP paper mentioned below.


If you use some of these resources in your research work, please cite the following paper:

author = {Adel, Heike and Sch\"utze, Hinrich},
title = {Using Mined Coreference Chains as a Resource for a Semantic Task},
booktitle={Proceedings of EMNLP},


[1] LDC2012T21, Agence France Presse 2010
[2] Heeyoung Lee, Yves Peirsman, Angel Chang, Nathanael Chambers, Mihai Surdeanu and Dan Jurafsky: "Stanford's multi-pass sieve coreference resolution system at the CoNLL-2011 shared task", 15th conference on CoNLL: Shared Task, pages 28-34, 2011.

Contact: Heike Adel (cis page)