A system for coreference resolution for the clinical narrative
- Jiaping Zheng1,
- Wendy W Chapman2,
- Timothy A Miller1,
- Chen Lin1,
- Rebecca S Crowley3,
- Guergana K Savova1
- 1Children's Hospital Boston and Harvard Medical School, Boston, Massachusetts, USA
- 2Division of Biomedical Informatics, University of California San Diego, San Diego, California, USA
- 3Department of Biomedical Informatics, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
- Correspondence to Dr Guergana K Savova, Children's Hospital Boston Informatics Program, Harvard Medical School, 300 Longwood Avenue, Boston, MA 02114, USA;
Contributors All authors of this manuscript contributed to (1) conception and design, acquisition of data, or analysis and interpretation of data, (2) drafting the article or revising it critically for important intellectual content, and (3) final approval of the version published.
- Received 15 September 2011
- Accepted 2 January 2012
- Published Online First 31 January 2012
Objective To research computational methods for coreference resolution in the clinical narrative and build a system implementing the best methods.
Methods The Ontology Development and Information Extraction corpus annotated for coreference relations consists of 7214 coreferential markables, forming 5992 pairs and 1304 chains. We trained classifiers with semantic, syntactic, and surface features pruned by feature selection. For the three system components—for the resolution of relative pronouns, personal pronouns, and noun phrases—we experimented with support vector machines with linear and radial basis function (RBF) kernels, decision trees, and perceptrons. Evaluation of algorithms and varied feature sets was performed using standard metrics.
Results The best performing combination is support vector machines with an RBF kernel and all features (MUC score=0.352, B3=0.690, CEAF=0.486, BLANC=0.596) outperforming a traditional decision tree baseline.
Discussion The application showed good performance similar to performance on general English text. The main error source was sentence distances exceeding a window of 10 sentences between markables. A possible solution to this problem is hinted at by the fact that coreferent markables sometimes occurred in predictable (although distant) note sections. Another system limitation is failure to fully utilize synonymy and ontological knowledge. Future work will investigate additional ways to incorporate syntactic features into the coreference problem.
Conclusion We investigated computational methods for coreference resolution in the clinical narrative. The best methods are released as modules of the open source Clinical Text Analysis and Knowledge Extraction System and Ontology Development and Information Extraction platforms.
- Coreference resolution
- natural language processing
- biomedical informatics
- information extraction
- machine learning
- human-computer interaction and human-centered computing
- Intelligent tutoring and tailored information representation
- Improving the education and skills training of health professionals
- providing just-in-time access to the biomedical literature and other health information
- applications that link biomedical knowledge from diverse primary sources (includes automated indexing)
- linking the genotype and phenotype
The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or ONC.
Funding This work was funded by NIH grant R01 CA127979 (PI Crowley), RC1LM010608 (PI Savova), SHARP award 90TR0002 (PI Chute), and U54LM008748 (PI Kohane).
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.