The biomedical discourse relation bank
2011

Biomedical Discourse Relation Bank

Sample size: 24 publication 10 minutes Evidence: moderate

Author Information

Author(s): Prasad Rashmi, McRoy Susan, Frid Nadya, Joshi Aravind, Yu Hong

Primary Institution: University of Pennsylvania

Hypothesis

Can discourse relations be reliably annotated in biomedical text?

Conclusion

Discourse relations can be reliably annotated in biomedical text, but more refined sense classification requires richer features or more annotated data.

Supporting Evidence

  • Annotated 24 open-access full-text biomedical articles from the GENIA corpus.
  • Achieved reliable inter-annotator agreement of over 80% for all sub-tasks.
  • Classifier performance stabilized at about 1900 training instances.
  • Classifier trained on PDTB performed poorly on BioDRB.

Takeaway

The study created a special bank to help understand how sentences in biomedical texts relate to each other, which can help in analyzing medical literature better.

Methodology

Annotated explicit and implicit discourse relations in 24 biomedical articles using adapted guidelines from the Penn Discourse TreeBank.

Limitations

The study did not annotate implicit or AltLex relations between events and situations within a single sentence.

Participant Demographics

Biomedical literature articles from the GENIA corpus.

Statistical Information

P-Value

0.05

Statistical Significance

p<0.05

Digital Object Identifier (DOI)

10.1186/1471-2105-12-188

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication