Skip to main content
. Author manuscript; available in PMC: 2016 Oct 14.
Published in final edited form as: J Mach Learn Res. 2016;17:132.

Table 5.

Averages (and standard deviations) of the proportion of the top-3 sentences extracted via the respective models from 50 prospectively annotated articles that were deemed relevant or best available by an annotator. The annotator was blinded to which model selected which sentences.

Method Precision@3
Population
Nguyen 0.907 (0.222)
SDS 0.927 (0.214)
Interventions
Nguyen 0.854 (0.254)
SDS 0.903 (0.245)
Outcomes
Nguyen 0.880 (0.208)
SDS 0.887 (0.196)