Elamur Research⌕
☰
home / papers / 2014 / Augmenting FrameNet Via PPDB

Augmenting FrameNet Via PPDB

Summary

FrameNet, the lexico-semantic resource based on frame semantics, has limited lexical coverage. Many words that should trigger a frame are missing, so parsers such as SEMAFOR need test-time heuristics for out-of-vocabulary items. Rastogi and Van Durme instead enlarge the training data. They take lexical rules from PPDB, The Paraphrase Database, and substitute paraphrases for single-word targets in FrameNet's fulltext annotated sentences. A 5-gram Kneser-Ney language model, built with KenLM on Annotated Gigaword, ranks the rewritten sentences.

They ran two setups. The first queried PPDB directly with each target word. The second first pooled lexical units from frames connected by inheritance, perspective or subframe relations, which yields far more candidates. They also measured what PPDB adds beyond WordNet synonyms. In a manual check of 25 paraphrase sets, the top three PPDB XL rewrites were grammatical and evoked the correct frame about 65% of the time, while roughly tripling coverage. A parser could therefore retrain on the expanded data instead of relying on OOV-handling tricks.

Key points

  • The input was FrameNet v1.5 (1,019 frames), whose fulltext portion has 4,026 sentences with 23,921 targets. Of these, 23,226 single-token targets were paraphrased.
  • Pooling lexical units across related frames (Experiment 2) raises the average lexical units per frame from 20.24 in FrameNet to 214.99 with PPDB XL.
  • After removing WordNet v3.0 synonyms, Experiment 1 still produced 283,126 new paraphrases with PPDB XL. Experiment 2 produced about 2.35 million with PPDB L.
  • With PPDB XL and only the top three language-model-ranked outputs kept, 65.27% of rewrites were fully correct (right frame and grammatical), and 87.5% evoked the correct frame.
  • The main errors came from single-token triggers that belong to multi-word expressions and from noisy PPDB entries. The generated corpus and the manual annotations were publicly released.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages

Topics: Semantic role labeling

Cited by

8 papers of our library cite it (of the 140673 whose bibliographies we read); 3 of them have a page here, 5 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.3115/v1/w14-2901

@inproceedings{rastogi2014augmenting,
  title = {{Augmenting FrameNet Via PPDB}},
  author = {Rastogi, Pushpendre and Durme, Benjamin Van},
  booktitle = {Proceedings of the Second Workshop on EVENTS: Definition, Detection, Coreference, and Representation},
  year = {2014},
  doi = {10.3115/v1/w14-2901},
  url = {https://doi.org/10.3115/v1/w14-2901}
}

References

18 references, 8 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Roni Ben Aharon, Idan Szpektor, and Ido Dagan. 2010. Generating entailment rules from framenet. In Proceedings of the ACL 2010 Conference Short Papers, pages 241–246. find our card
  2. Collin Baker, Michael Ellsworth, and Katrin Erk. 2007. Semeval’07 task 19: frame semantic structure extraction. In Proceedings of the 4th International Workshop on Semantic Evaluations, pages 99–104. ACL. find
  3. Bob Coyne and Owen Rambow. 2009. Lexpar: A freely available english paraphrase lexicon automatically extracted from framenet. 2012 IEEE Sixth International Conference on Semantic Computing, pages 53–58. find
  4. Dipanjan Das, Desai Chen, André F. T. Martins, Nathan Schneider, and Noah A. Smith. 2013. Frame-semantic parsing. Computational Linguistics, 40(1):9–56. find our card
  5. Oscar Ferrández, Michael Ellsworth, Rafael Munoz, and Collin F Baker. 2010. Aligning framenet and wordnet based on semantic neighborhoods. In LREC, volume 10, pages 310–314. find our card
  6. Charles J Fillmore and Collin F Baker. 2001. Frame semantics for text understanding. In Proceedings of WordNet and Other Lexical Resources Workshop, NAACL. find
  7. Marco Fossati, Claudio Giuliano, and Sara Tonelli. 2013. Outsourcing framenet to the crowd. In Proceedings of the 51st Annual Meeting of the ACL, pages 742–747. find our card
  8. Juri Ganitkevitch, Benjamin Van Durme, and Chris Callison-Burch. 2013. PPDB: The paraphrase database. In Proceedings of NAACL-HLT, pages 758–764, Atlanta, Georgia, June. ACL. find our card
  9. Daniel Gildea and Daniel Jurafsky. 2002. Automatic labeling of semantic roles. Computational linguistics, 28(3):245–288. find our card
  10. Kenneth Heafield, Ivan Pouzyrevsky, Jonathan H. Clark, and Philipp Koehn. 2013. Scalable modified Kneser-Ney language model estimation. In Proceedings of the 51st Annual Meeting of the ACL, Sofia, Bulgaria. find our card
  11. Jiřı́ Materna. 2012. Lda-frames: An unsupervised approach to generating semantic frames. In Computational Linguistics and Intelligent Text Processing, volume 7181 of Lecture Notes in Computer Science, pages 376–387. Springer Berlin Heidelberg. find
  12. Courtney Napoles, Matthew Gormley, and Benjamin Van Durme. 2012. Annotated gigaword. In Proceedings of the Joint Workshop on Automatic Knowledge Base Construction and Web-scale Knowledge Extraction, pages 95–100. ACL. find
  13. Brendan O’ Connor. 2012. Learning frames from text with an unsupervised latent variable model. In Technical Report. Carnegie Mellon University. find
  14. Jansen Orfan and James Allen. 2013. Toward learning high-level semantic frames from definitions. In Proceedings of the Second Annual Conference on Advances in Cognitive Systems, volume 125. find
  15. Martha Palmer, Daniel Gildea, and Paul Kingsbury. 2005. The proposition bank: An annotated corpus of semantic roles. Computational Linguistics, 31(1):71–106. find our card
  16. Rion Snow, Daniel Jurafsky, and Andrew Y Ng. 2004. Learning syntactic patterns for automatic hypernym discovery. In NIPS, volume 17, pages 1297–1304. find
  17. Christopher Walker, Stephanie Strassel, Julie Medero, and Kazuaki Maeda. 2006. find
  18. Ace 2005 multilingual training corpus. Linguistic Data Consortium, Philadelphia. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/papers/rastogi-2014-augmenting-framenet-ppdb/