Skip to main content
. Author manuscript; available in PMC: 2016 Nov 14.
Published in final edited form as: KDD. 2016 Aug;2016:855–864. doi: 10.1145/2939672.2939754

Figure 4.

Figure 4

Performance evaluation of different benchmarks on varying the amount of labeled data used for training. The x axis denotes the fraction of labeled data, whereas the y axis in the top and bottom rows denote the Micro-F1 and Macro-F1 scores respectively. DeepWalk and node2vec give comparable performance on PPL In all other networks, across all fractions of labeled data node2vec performs best.