SPred is an approach for large-scale harvesting of Semantic Predicates developed by Tiziano Flati and Roberto Navigli.
Datasets
We are releasing a package that mainly contains:
3 text files containing the list of predicates used as input to the system
semantic predicates for 2 different sets of lexical predicates (namely, OXFORD and K&H)
3 evaluation datasets used in the experimental setup; 2 refer to the semantic class ranking evaluation (Oxford's and K&H's predicates) and one refers to the Oxford argument disambiguation evaluation
Predicate archives contain the semantic class ranking for each of the semantic predicate.
The evaluation datasets contain items and the corresponding annotations provided.
Downloads
Complete Download
SPred complete archive: text files, semantic predicate archives, evaluation datasets, README (revision 1 August 2013).
Selective Download
README
README file containing detailed information about the text files, the semantic predicates and the datasets format (revision 1 August 2013).
When citing SPred and our experimental results, please refer to the following paper:
Tiziano Flati, Roberto Navigli. SPred: Large-scale Harvesting of Semantic Predicates. Proc. of the 51st Annual Meeting of the Association for Computational Linguistics (ACL 2013), Sofia, Bulgaria,
August 4-9, 2013
