Skip to main content
. 2016 Aug 4;11(8):e0159644. doi: 10.1371/journal.pone.0159644

Fig 1. The general architecture of our approach.

Fig 1

R: record; Pair R1 R2 and Pair R1 RN are expert labelled duplicate and distinct pairs respectively; Binary: whether a pair is duplicate or distinct; Multi: multiple duplicate types and distinct pairs; Ablation: quantify the impacts of different features; Error: quantify erroneous cases to characterise challenging cases; Generalisation: whether model can be applied to a different dataset.