Ruprecht-Karls-UniversitÃ<83>¤t Heidelberg

IndIA: Inducing Implicit Arguments

From this page you can download various resources described in the following publications:

  • Roth, M. and Frank, A. (2013):
    Automatically identifying implicit arguments to improve argument linking and coherence modeling. Proceedings of the Second Joint Conference on Lexical and Computational Semantics (*SEM), Atlanta, Georgia, USA, June, pp. 306-316, Association for Computational Linguistics. [ .pdf ]
  • Roth, M. and Frank, A. (2012):
    Aligning predicates across monolingual comparable texts using graph-based clustering. Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning (EMNLP), Jeju Island, Republic of Korea, July, pp. 171-182. [ .pdf ]
  • Roth, M. and Frank, A. (2012):
    Aligning predicate argument structures in monolingual comparable texts: A new corpus for a new task. Proceedings of the First Joint Conference on Lexical and Computational Semantics (*SEM), Montréal, Canada, June, pp. 218-227. [ .pdf ]

Overview

The download area contains four resources:

  • Document IDs of comparable texts extracted from the Gigaword corpus
  • Gold standard alignments between predicate argument structures across texts
  • Automatically computed high precision alignments for the full pairwise corpus
  • Induced instances of implicit arguments and their discourse antecedents

Download

Automatically extracted pairs of comparable texts from Gigaword

This is a sub-set of the English Gigaword Fifth Edition, comprising of pairwise documents that are predicted to be about the same events and entities.

You can download the corpus in the form of ID lists here.

Gold standard alignment annotations

This contains a set of predicate alignments labeled by human annotators.

You can download the gold standard as stand-off annotations here.

Automatically computed high precision alignments

This is a set of automatically computed high precision alignments for the full data set of text pairs described above.

You can download the automatic alignments here.

Implicit arguments and discourse antecedents

This is a set of implicit arguments that were automatically induced by comparing the argument structures of automatically aligned predicates across pairs of comparable texts.

You can download the induced instances of implicit arguments here.

Contact

Please feel free to send us your technical question, requests and bug reports to technik@cl.uni-heidelberg.de