any-language-frames

Multilingual datasets for the paper "Any-language frame-semantic parsing". The project contains three directories:

1. /annotated/ contains the directories with annotated files, named following a language_annotatorconvention.
2. /preannotated/ contains the pre-annotated files, with the FrameNet+BabelNet frame labels the annotators had to choose from.
3. /preprocessed/ contains the preprocessed files, with POS and dependency trees predicted using TreeTagger (Schmid,1995) and TurboParser (Martins et al, 2010) on Universal Dependencies v1.1.

If you use this resource, please cite the following article.

@InProceedings{johannsen-martinezalonso-sogaard:2015:EMNLP,
  author    = {Johannsen, Anders  and  Mart\'{i}nez Alonso, H\'{e}ctor  and  S{\o}gaard, Anders},
  title     = {Any-language frame-semantic parsing},
  booktitle = {Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing},
  month     = {September},
  year      = {2015},
  address   = {Lisbon, Portugal},
  publisher = {Association for Computational Linguistics},
  pages     = {2062--2066},
  url       = {http://aclweb.org/anthology/D15-1245}
}