Elamur ResearchβŒ•
☰
home / papers / 2007 / Mutaphrase: Paraphrasing with FrameNet

Mutaphrase: Paraphrasing with FrameNet

Summary

Statistical NLP systems benefit from more training data. One way to get it is to restate existing sentences so the meaning stays similar while the wording and word order change. Most earlier paraphrase generators learned rewrites from corpora of known paraphrase pairs. Mutaphrase is a preliminary system that instead builds paraphrases algorithmically from FrameNet, a hand-annotated lexico-semantic database that maps meaning to form. The input is a sentence with FrameNet annotation, currently done by hand. The system turns it into a syntactico-semantic tree and automatically extracts ordered valences, i.e. patterns of frame elements with phrase types and grammatical functions. It then recursively rewrites each frame node using alternative valences from the same lexical unit, from other units of the same frame, or from related frames. The output is a very large set of candidates that vary in both syntactic and semantic distance from the input. These are ranked only by how often each valence occurs in FrameNet, so some results are semantically odd. The approach matters because it offers controllable data expansion for language modeling, MT and generation without parallel paraphrase corpora.

Key points

  • For "I like eating cheese", the system produced 8403 paraphrases. The top-ranked one, "I resented drinking cheese", shows the problem of semantic mismatch. "Nausea seems a commonplace symptom" produced 502 paraphrases.
  • Each annotated LU typically has 2–10 valence patterns. These cover active/passive, the dative alternation and optional elements. Multiword LUs such as take off.v are also handled; its two orderings occur 57/68 and 7/68 times.
  • Long sentences can yield 10,000+ paraphrases but use a great deal of time and memory. The authors suggest earlier pruning as a fix.
  • Current limitations:
    • The input requires manual FrameNet annotation.
    • Conjunctions and relative clauses are poorly supported.
    • Nonsensical outputs such as sipping meat are not filtered.
  • Planned work includes:
    • automatic semantic parsing (e.g. Shalmaneser);
    • WordNet-based similarity filtering and synonym expansion;
    • scoring by syntactic edit distance and language models;
    • multilingual paraphrasing via FrameNet-like resources such as SALSA.

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

Fields: Linguistics & languages Β· Machine translation

Topics: Semantic role labeling

Cited by

14 papers of our library cite it (of the 140673 whose bibliographies we read); 1 of them have a page here, 13 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work.

@inproceedings{ellsworth2007mutaphrase,
  title = {{Mutaphrase: Paraphrasing with FrameNet}},
  author = {Ellsworth, Michael and Janin, Adam},
  booktitle = {Proceedings of the ACL-PASCAL Workshop on Textual Entailment and Paraphrasing},
  year = {2007},
  url = {https://aclanthology.org/W07-1424/}
}

References

27 references, 8 of them with a card of ours. β†— leads to the work, Β«findΒ» searches for one we have no id for yet.

  1. U. Baldewein, K. Erk, S. Padó, and D. Prescher. 2004. Semantic role labelling with similarity-based generalization using EM-based clustering. In R. Mihalcea and P. Edmonds, editors, Senseval-3: Third International Workshop on the Evaluation of Systems for the Semantic Analysis of Text, pages 64–68, Barcelona, Spain, July. Association for Computational Linguistics. find
  2. C. Bannard and C. Callison-Burch. 2005. Paraphrasing with bilingual parallel corpora. In Proceedings of the 43rd annual meeting of the Association for Computational Linguistics (ACL), pages 597–604, Ann Arbor, June. find our card
  3. R. Barzilay and L. Lee. 2003. Learning to paraphrase: An unsupervised approach using multiple-sequence alignment. In Proceedings of the Human Language Technology Conference (HLT), pages 16–23, Edmonton, Canada, May. β†— our card
  4. R. Barzilay and K. McKeown. 2001. Extracting paraphrases from a parallel corpus. In Proceedings of the Annual Meeting of the Association for Computational Linguistics (ACL), pages 50–57, Toulouse, July. find our card
  5. C. Callison-Burch, P. Koehn, and M. Osborne. 2006. Improved statistical machine translation using paraphrases. In Proceedings of the Human Language Technology Conference (HLT), pages 17–24, New York City, June. find
  6. W. Dolan, C. Quirk, and C. Brockett. 2004. Unsupervised construction of large paraphrase corpora: Exploiting massively parallel news sources. In Proceedings of the 20th International Conference on Computational Linguistics (COLING), Geneva, Switzerland, August. find our card
  7. A.C. Doyle. 1902. Hound of the Baskervilles. Project Gutenburg web site. http://www.gutenberg.org/dirs/etext02/bskrv11a.txt. β†—
  8. K. Erk and S. Padó. 2006. Shalmaneser β€” a flexible toolbox for semantic role assignment. In Proceedings of the Fifth International Conference on Language Resources and Evaluation (LREC), pages 527– 532, Genoa, Italy, May. find
  9. C. Fellbaum. 1998. WordNet: An Electronic Lexical Database. The MIT Press, May. find
  10. T. Fontenelle, editor. 2003. International Journal of Lexicography Special Issue on FrameNet and Frame Semantics. Oxford University Press, September. volume 16(3). find
  11. FrameNet. 2007a. The FrameNet task on SemEval web site. http://nlp.cs.swarthmore.edu/semeval/tasks/task19/summary.shtml. β†—
  12. FrameNet. 2007b. FrameNet web site. http://framenet.icsi.berkeley.edu. β†—
  13. Japanese FrameNet. 2007c. Japanese FrameNet web site. http://jfn.st.hc.keio.ac.jp/. β†—
  14. Spanish FrameNet. 2007d. Spanish FrameNet web site. http://gemini.uab.es:9080/SFNsite. β†—
  15. A. Ibrahim, B. Katz, and J. Lin. 2003. Extracting structural paraphrases from aligned monolingual corpora. In Proceedings of the Second International Workshop on Paraphrasing, pages 57–64, Sapporo, Japan, July. β†—
  16. D. Kauchak and R. Barzilay. 2006. Paraphrasing for automatic evaluation. In Proceedings of the Human Language Technology Conference (HLT), pages 455– 462, New York City, June. find our card
  17. R. Kittredge. 2002. Paraphrasing for condensation in journal abstracting. Journal of Biomedical Informatics, 35(4):265–277. β†—
  18. D. Lin and P. Pantel. 2001. Discovery of inference rules for question-answering. Natural Language Engineering, 7(4):343–360. β†—
  19. K. Litkowski. 2004. Senseval-3 task: Automatic labeling of semantic roles. In R. Mihalcea and P. Edmonds, editors, Senseval-3: Third International Workshop on the Evaluation of Systems for the Semantic Analysis of Text, pages 9–12, Barcelona, Spain, July. Association for Computational Linguistics. find our card
  20. D. Moldovan, R. GΔ±Μ‚rju, M. Olteanu, and O. Fortu. 2004. SVM classification of FrameNet semantic roles. In Rada Mihalcea and Phil Edmonds, editors, Senseval-3: Third International Workshop on the Evaluation of Systems for the Semantic Analysis of Text, pages 167– 170, Barcelona, Spain, July. Association for Computational Linguistics. find
  21. B. Pang, K. Knight, and D. Marcu. 2003. Syntax-based alignment of multiple translations: Extracting paraphrases and generating new sentences. In Proceedings of the Human Language Technology Conference (HLT), pages 102–109, Edmonton, Canada, May. find our card
  22. C. Quirk, C. Brockett, and W. Dolan. 2004. Monolingual machine translation for paraphrase generation. In Proceedings of the 2004 Conference on Empirical Methods in Natural Language Processing, pages 142–149, Barcelona Spain, July. find our card
  23. F. Rinaldi, J. Dowdall, K. Kaljurand, M. Hess, and D. Mollá. 2003. Exploiting paraphrases in a question answering system. In Proceedings of the Second International Workshop on Paraphrasing, pages 25–32, July. find
  24. SALSA. 2007. SALSA Project web site. http://www.coli.uni-saarland.de/projects/salsa/. β†—
  25. Y. Shinyama, S. Sekine, K. Sudo, and R. Grishman. 2002. Automatic paraphrase acquisition from news articles. In Proceedings of Human Language Technology Conference (HLT), pages 40–46, San Diego, March. find
  26. WordNet. 2006. WordNet web site. http://wordnet.princeton.edu. β†—
  27. I. Zukerman and B. Raskutti. 2002. Lexical query paraphrasing for document retrieval. In Proceedings of the 19th International Conference on Computational Linguistics (COLING), pages 1–7, Taipei, Taiwan, August. find

↑ back to top

This page in Markdown β†’ https://md.research.elamur.ai/free/papers/ellsworth-2007-mutaphrase-paraphrasing-framenet/