Ontology-based Interactive Information Extraction from Scientific Abstracts
Author Information
Author(s): David Milward, Marcus Bjareland, William Hayes, Michelle Maxwell, Lisa Oberg, Nick Tilford, James Thomas, Roger Hale, Sylvia Knight, Julie Barnes
Primary Institution: Linguamatics Ltd
Hypothesis
Can an ontology-based interactive information extraction system improve the extraction of co-factors from scientific literature?
Conclusion
The OBIIE system effectively retrieves co-factors from scientific abstracts, achieving high recall and saving time compared to manual methods.
Supporting Evidence
- The OBIIE system retrieved co-factors that manual curation missed.
- Recall for the OBIIE system was 90% for both AR and LXR.
- The system allows users to refine queries for more precise results.
- Using the OBIIE system saved significant time compared to manual extraction.
- The study highlights the effectiveness of using ontologies in information extraction.
Takeaway
This study shows a new tool that helps scientists quickly find important information from lots of scientific papers, making it easier to discover new facts.
Methodology
The study compared manual extraction of co-factors from abstracts with the use of an ontology-based interactive information extraction system.
Limitations
The study did not achieve 100% recall due to missing patterns and references spanning multiple sentences.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website