Elamur Research⌕
☰
home / papers / 2010 / Memory-Based Resolution of In-Sentence Scopes of Hedge Cues

Memory-Based Resolution of In-Sentence Scopes of Hedge Cues

Summary

The paper describes the system that CLiPS (University of Antwerp) submitted for the closed-track competition of the CoNLL-2010 Shared Task on hedge detection. The task has two parts: deciding whether a sentence carries uncertain information (Task 1), and identifying hedge cues and the words each cue governs in biomedical text (Task 2).

For Wikipedia data, an SVM labels every token using a hand-picked set of frequent weasel cues, and the sentence label is derived from those token decisions. For the biological data (BIO-ABS, BIO-ART), the authors chain two memory-based classifiers. The first tags cue tokens in a begin/inside/outside scheme, so multiword cues can be found. The second runs once per cue and marks the first and last tokens of its scope. It draws on shallow syntax and dependency syntax from the GDep parser, plus a cue lexicon built from the training data. Postprocessing then handles references to bibliography, tables and figures.

The scope resolver obtained the top F1 in Task 2 (57.32). Even so, the score shows that hedge scope resolution in full biomedical articles remains hard. The main error sources are mistakes in cue detection, flawed dependency parses and metalinguistic sentences.

Key points

  • Task 2 result: best F1 among participants, 57.32. Without the reference and scope postprocessing steps, F1 falls to 50.02 and 46.52 respectively.
  • Preprocessing tools:
    • WIKI data was parsed with MBSP (lemmas, PoS, chunks) and MaltParser (dependencies).
    • BIO data was parsed with GDep, which was trained on abstracts and is therefore weaker on full articles.
    • Converting the gold xml to CoNLL format kept scorer F1 between 99.10 and 100.00.
  • WIKI training data: it contains 3133 weasels across 1984 types, and 63% of the types are unique. For this reason the authors selected 42 frequent weasel cue tokens by hand rather than trying to detect weasels directly.
  • Differences from Morante and Daelemans (2009):
    • one classifier replaces three classifiers plus a metalearner;
    • dependency features are added to shallow ones;
    • a cue lexicon built from the training data is added.
  • Difficult inputs: ambiguous cues such as "or", subordinate clauses that wrongly end up inside a scope, and a test article about hedge annotation itself, whose metalanguage produced false positives.

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

12 papers of our library cite it (of the 140673 whose bibliographies we read); 1 of them have a page here, 11 not yet.

Cite

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

@inproceedings{morante2010memory,
  title = {{Memory-Based Resolution of In-Sentence Scopes of Hedge Cues}},
  author = {Morante, Roser and Asch, Vincent Van and Daelemans, Walter},
  booktitle = {Proceedings of the Fourteenth Conference on Computational Natural Language Learning – Shared Task},
  year = {2010},
  url = {https://aclanthology.org/W10-3006/}
}

References

22 references, 7 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Sabine Buchholz and Erwin Marsi. 2006. CoNLL-X shared task on multilingual dependency parsing. In Proceedings of the CoNLL-X Shared Task, New York. SIGNLL. find our card
  2. Thomas M. Cover and Peter E. Hart. 1967. Nearest neighbor pattern classification. Institute of Electrical and Electronics Engineers Transactions on Information Theory, 13:21–27. ↗
  3. Walter Daelemans and Antal van den Bosch. 2005. Memory-based language processing. Cambridge University Press, Cambridge, UK. ↗
  4. Walter Daelemans, Jakub Zavrel, Ko Van der Sloot, and Antal Van den Bosch. 2009. TiMBL: Tilburg Memory Based Learner, version 6.2, Reference Guide. Number 09-01 in Technical Report Series. Tilburg, The Netherlands. find
  5. Chrysanne Di Marco and Robert 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. Richárd Farkas, Veronika Vincze, György Móra, János Csirik, and György Szarvas. 2010. The CoNLL-2010 Shared Task: Learning to Detect Hedges and their Scope in Natural Language Text. In Proceedings of the Fourteenth Conference on Computational Natural Language Learning (CoNLL-2010): Shared Task, pages 1–12, Uppsala, Sweden, July. Association for Computational Linguistics. find our card
  7. Carol Friedman, Philip Alderson, John Austin, James J. Cimino, and Stephen B. Johnson. 1994. A general natural–language text processor for clinical radiology. Journal of the American Medical Informatics Association, 1(2):161–174. ↗
  8. Ken Hyland. 1998. Hedging in scientific research articles. John Benjamins B.V, Amsterdam. ↗ our card
  9. Thorsten Joachims. 2002. Learning to Classify Text Using Support Vector Machines, volume 668 of The Springer International Series in Engineering and Computer Science. Springer. find
  10. Halil Kilicoglu and Sabine Bergler. 2008. Recognizing speculative language in biomedical research articles: a linguistically motivated perspective. BMC Bioinformatics, 9(Suppl 11):S10. find
  11. George Lakoff. 1972. Hedges: a study in meaning criteria and the logic of fuzzy concepts. Chicago Linguistics Society Papers, 8:183–228. ↗ our card
  12. Marc Light, Xin Y.Qiu, and Padmini Srinivasan. 2004. The language of bioscience: facts, speculations, and statements in between. In Proceedings of the BioLINK 2004, pages 17–24. find
  13. Ben Medlock and Ted Briscoe. 2007. Weakly supervised learning for hedge classification in scientific literature. In Proceedings of ACL 2007, pages 992– 999. find our card
  14. Ben Medlock. 2008. Exploring hedge identification in biomedical literature. Journal of Biomedical Informatics, 41:636–654. ↗
  15. Roser Morante and Walter Daelemans. 2009. Learning the scope of hedge cues in biomedical texts. In Proceedings of BioNLP 2009, pages 28–36, Boulder, Colorado. find our card
  16. Joakim Nivre. 2006. Inductive Dependency Parsing, volume 34 of Text, Speech and Language Technology. Springer. find
  17. Arzucan Özgür and Dragomir R. Radev. 2009. Detecting speculations and their scopes in scientific text. In Proceedings of EMNLP 2009, pages 1398–1407, Singapore. find
  18. Frank R. Palmer. 1986. Mood and modality. CUP, Cambridge, UK. find
  19. Kenji Sagae and Jun’ichi Tsujii. 2007. Dependency parsing and domain adaptation with LR models and parser ensembles. In Proceedings of CoNLL 2007: Shared Task, pages 82–94, Prague, Czech Republic. find
  20. György Szarvas. 2008. Hedge classification in biomedical texts with a weakly supervised selection of keywords. In Proceedings of ACL 2008, pages 281–289, Columbus, Ohio, USA. ACL. find our card
  21. Paul Thompson, Giulia Venturi, John McNaught, Simonetta Montemagni, and Sophia Ananiadou. 2008. Categorising modality in biomedical texts. In Proceedings of the LREC 2008 Workshop on Building and Evaluating Resources for Biomedical Text Mining 2008, pages 27–34, Marrakech. LREC. find
  22. Veronika Vincze, György Szarvas, Richárd Farkas, György Móra, and János Csirik. 2008. The Bio-Scope corpus: biomedical texts annotated for uncertainty, negation and their scopes. BMC Bioinformatics, 9(Suppl 11):S9. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/morante-2010-memory-based-resolution-sentence-scopes/