Recognizing Speculative Language in Biomedical Research Articles
Author Information
Author(s): Kilicoglu Halil, Bergler Sabine
Primary Institution: Department of Computer Science and Software Engineering, Concordia University
Hypothesis
Can a linguistically motivated approach effectively recognize speculative language in biomedical research articles?
Conclusion
The study shows that speculative language can be recognized successfully using a linguistically motivated approach, and that the selection of hedging devices affects the speculative strength of sentences.
Supporting Evidence
- The system achieved a precision-recall breakeven point of 0.85 on the fruit-fly dataset.
- Using semi-automatic weighting, the system improved the BEP on the BMC dataset to 0.82.
- The study's approach is competitive with previously reported best results in recognizing hedging.
Takeaway
This study helps computers understand when scientists are being unsure or tentative in their writing, which is important for accurately interpreting research.
Methodology
The study used two publicly available hedging datasets to evaluate a system that recognizes hedging cues and assigns weights to them based on their speculative strength.
Potential Biases
The reliance on specific datasets may introduce bias in recognizing hedging across different contexts.
Limitations
The training set was small, which may limit the generalizability of the findings.
Statistical Information
P-Value
<0.01
Statistical Significance
p<0.01
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website