Gene Mention Normalization and Interaction Extraction
Author Information
Author(s): Hakenberg Jörg, Plake Conrad, Royer Loic, Strobelt Hendrik, Leser Ulf, Schroeder Michael
Primary Institution: Technische Universität Dresden
Hypothesis
Can gene mention normalization and protein-protein interaction extraction be improved using context models and sentence motifs?
Conclusion
The proposed methods for gene mention normalization and protein-protein interaction extraction are fully automated and perform comparably to systems requiring human intervention.
Supporting Evidence
- The gene mention normalization method achieved an f-measure of 86.4%.
- The protein-protein interaction extraction method achieved an f-measure of 24.4%.
- Using background knowledge significantly improved the precision of gene identification.
Takeaway
This study helps computers understand gene names and how proteins interact by using special patterns and context from scientific texts.
Methodology
The study used context models to improve gene mention normalization and sentence motifs for extracting protein-protein interactions.
Limitations
The methods may struggle with ambiguous gene names and require high-quality background knowledge.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website