Fig. 8From: SURREAL: Subgraph Robust Representation LearningPerformance comparison of SURREAL using two input graphs: entire graph and supergraph. The experiment is conducted using Wikipedia dataset. The |NR(u)| is varied from 200 to 1,200, while |NE(u)|=1,200 and d=128. The performance degrades when using the supergraph instead of entire graph as an input graphBack to article page