Figure 8

Comparison of classification accuracy for two different training data sets (training set 1 was extracted from concept network snapshots ${\mathit{G}}_{{\mathit{t}}_{\mathbf{1}}}\mathbf{=}\mathbf{1991}\mathbf{-}\mathbf{1995}$ and ${\mathit{G}}_{{\mathit{t}}_2}\mathbf{=}\mathbf{1996}\mathbf{-}\mathbf{2000}$; training data set 2 was extracted from concept network snapshots $G_{t_2}$ and ${\mathit{G}}_{{\mathit{t}}_3}\mathbf{=}\mathbf{2001}\mathbf{-}\mathbf{2005}$).