Graph Based Multi-View Learning for CDL Relation Classification

Haibo Li, Yutaka Matsuo, Mitsuru Ishizuka

Abstract: To understand text contents better, many research efforts have been made exploring detection and classification of the semantic relation between a concept pair. As described herein, we present our study of a semantic relation classification task as a graph-based multi-view learning task: each intra-view graph is constructed with instances in the view; a node's label ldquoscorerdquo is propagated on each intra-view graph and inter-view graph. This combination of multi-view learning and graph-based method can reduce the influence from violation of a background assumption of multi-view learning algorithms-view compatibility. The proposed algorithm is evaluated using the Concept Description Language for Natural Language (CDL.nl) corpus. The experiment results validate its effectiveness.