Elamur Research
☰
home / papers / 2009 / Learning the Scope of Hedge Cues in Biomedical Texts

Learning the Scope of Hedge Cues in Biomedical Texts

Written for the Research library (claude-opus-5-5) from the paper text, in our own words: the license "CC BY-NC-SA 3.0 (ACL Anthology, before 2016)" does not allow redistributing the text itself. Original — aclanthology.org/W09-1304.pdf

Summary

Treating speculative statements in biomedical literature as established facts can corrupt information extraction. Earlier machine learning work only labelled whole sentences as speculative or definite. Morante and Daelemans go further: they identify which words a given hedge cue actually affects. They adapt their earlier negation scope system to hedging and split the problem into two token classification steps. First, B/I/O tagging detects cues, including multiword ones, such as may, suggest or or. Second, for each cue, tokens are labelled as the first or last element of its scope, using a metalearning combination of classifiers. Features draw on GENIA tagger output (lemmas, POS, chunk tags) and surrounding context. Experiments use the three BioScope subcorpora: models are trained on abstracts and also applied to full papers and clinical reports. On abstracts, cue detection reaches 84.77% F1, and scope finding with predicted cues gets 65.55% PCS, nearly matching the negation system. Results fall sharply on the other text types, which shows that hedge detection depends strongly on domain. The authors present this as the first system that learns hedge scope.

Key points

  • Data: BioScope corpus (clinical radiology reports, biological full papers, GENIA abstracts). Hedged sentences make up 13.39%, 19.44% and 17.70% of these subcorpora, respectively.
  • BioScope contains 176 distinct hedge cues versus 38 negation cues, and only 15.90% of hedge cues occur in all three subcorpora.
  • Cue identification uses IGTREE with token features and a window of 3 tokens on each side. It reaches 84.77% F1 on abstracts but only 41.92% F1 on clinical texts.
  • Scope finding with predicted cues reaches 65.55 / 35.92 / 26.21 PCS on abstracts / papers / clinical, compared with 66.07% PCS for the negation system on abstracts.
  • Cues that are absent from the abstracts training data score near 0 PCS on other subcorpora. This confirms Szarvas (2008), who found that hedge classifiers transfer poorly across domains.

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

Cited by

55 papers of our library cite it (of the 140673 whose bibliographies we read); 2 of them have a page here, 53 not yet.

Cite

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

@inproceedings{morante2009learning,
  title = {{Learning the Scope of Hedge Cues in Biomedical Texts}},
  author = {Morante, Roser and Daelemans, Walter},
  booktitle = {Proceedings of the BioNLP 2009 Workshop},
  year = {2009},
  doi = {10.3115/1572364.1572369},
  url = {https://doi.org/10.3115/1572364.1572369}
}

References

28 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. S. Buchholz and E. Marsi. 2006. CoNLL-X shared task on multilingual dependency parsing. In Proc. of the X CoNLL Shared Task, New York. SIGNLL. find our card
  2. N. Collier, H.S. Park, N. Ogata, Y. Tateisi, C. Nobata, T. Sekimizu, H. Imai, and J. Tsujii. 1999. The GE-NIA project: corpus-based knowledge acquisition and information extraction from genome research papers. In Proc. of EACL 1999. find
  3. T. M. Cover and P. E. Hart. 1967. Nearest neighbor pattern classification. Institute of Electrical and Electronics Engineers Transactions on Information Theory, 13:21–27. find
  4. W. Daelemans, J. Zavrel, K. Van der Sloot, and A. Van den Bosch. 2007. TiMBL: Tilburg memory based learner, version 6.1, reference guide. Technical Report Series 07-07, ILK, Tilburg, The Netherlands. find
  5. C. Di Marco and R.E. Mercer, 2005. Computing attitude and affect in text: Theory and applications, chapter Hedging in scientific articles as a means of classifying citations. Springer-Verlag, Dordrecht. find
  6. C. Friedman, P. Alderson, J. Austin, J.J. Cimino, and S.B. Johnson. 1994. A general natural–language text processor for clinical radiology. JAMIA, 1(2):161–174. find
  7. K. Hyland. 1998. Hedging in scientific research articles. John Benjamins B.V, Amsterdam. find our card
  8. T. Joachims, 1999. Advances in Kernel Methods - Support Vector Learning, chapter Making large-Scale SVM Learning Practical, pages 169–184. MIT-Press, Cambridge, MA. find
  9. H. Kilicoglu and S. Bergler. 2008. Recognizing speculative language in biomedical research articles: a linguistically motivated perspective. BMC Bioinformatics, 9(Suppl 11):S10. find
  10. M. Krallinger, F. Leitner, C. Rodriguez-Penagos, and A. Valencia. 2008a. Overview of the protein–protein interaction annotation extraction task of BioCreative II. Genome Biology, 9(Suppl 2):S4. find
  11. M. Krallinger, A. Valencia, and L. Hirschman. 2008b. Linking genes to literature: text mining, information extraction, and retrieval applications for biology. Genome Biology, 9(Suppl 2):S8. find
  12. M. Krauthammer, P. Kra, I. Iossifov, S.M. Gomez, G. Hripcsak, V. Hatzivassiloglou, C. Friedman, and A.Rzhetsky. 2002. Of truth and pathways: chasing bits of information through myriads of articles. Bioinformatics, 18(Suppl 1):S249–57. find
  13. J. Lafferty, A. McCallum, and F. Pereira. 2001. Conditional random fields: Probabilistic models for segmenting and labeling sequence data. In Proc. of ICML 2001, pages 282–289. find
  14. G. Lakoff. 1972. Hedges: a study in meaning criteria and the logic of fuzzy concepts. Chicago Linguistics Society Papers, 8:183–228. find our card
  15. M. Light, X.Y.Qiu, and P. Srinivasan. 2004. The language of bioscience: facts, speculations, and statements in between. In Proc. of the BioLINK 2004, pages 17–24. find
  16. B. Medlock and T. Briscoe. 2007. Weakly supervised learning for hedge classification in scientific literature. In Proc. of ACL 2007, pages 992–999. find our card
  17. B. Medlock. 2008. Exploring hedge identification in biomedical literature. JBI, 41:636–654. find
  18. T. Mitsumori, M. Murata, Y. Fukuda, K Doi, and H. Doi. 2006. Extracting protein-protein interaction information from biomedical text with svm. IEICE - Trans. Inf. Syst., E89-D(8):2464–2466. find
  19. R. Morante and W. Daelemans. 2009. A metalearning approach to processing the scope of negation. In Proc. of CoNLL 2009, Boulder, Colorado. find
  20. F.R. Palmer. 1986. Mood and modality. CUP, Cambridge, UK. find
  21. R. Saurı́, M. Verhagen, and J. Pustejovsky. 2006. Annotating and recognizing event modality in text. In Proc. of FLAIRS 2006, pages 333–339. find
  22. G. Szarvas, V. Vincze, R. Farkas, and J. Csirik. 2008. The BioScope corpus: annotation for negation, uncertainty and their scope in biomedical texts. In Proc. of BioNLP 2008, pages 38–45, Columbus, Ohio. ACL. find
  23. G. Szarvas. 2008. Hedge classification in biomedical texts with a weakly supervised selection of keywords. In Proc. of ACL 2008, pages 281–289, Columbus, Ohio, USA. ACL. find
  24. P. Thompson, G. Venturi, J. McNaught, S. Montemagni, and S. Ananiadou. 2008. Categorising modality in biomedical texts. In Proc. of the LREC 2008 Workshop on Building and Evaluating Resources for Biomedical Text Mining 2008, pages 27–34, Marrakech. LREC. find
  25. Y. Tsuruoka and J. Tsujii. 2005. Bidirectional inference with the easiest-first strategy for tagging sequence data. In Proc. of HLT/EMNLP 2005, pages 467–474. find
  26. Y. Tsuruoka, Y. Tateishi, J. Kim, T. Ohta, J. McNaught, S. Ananiadou, and J. Tsujii, 2005. Advances in Informatics - 10th Panhellenic Conference on Informatics, volume 3746 of LNCS, chapter Part-of-Speech Tagger for Biomedical Text, Advances in Informatics, pages 382–392. Springer, Berlin/Heidelberg. find
  27. C.J. Van Rijsbergen. 1979. Information Retrieval. Butterworths, London. find
  28. V. Vincze, G. Szarvas, R. Farkas, G. Móra, and J. Csirik. 2008. The BioScope corpus: biomedical texts annotated for uncertainty, negation and their scopes. BMC Bioinformatics, 9(Suppl 11):S9. find