We propose a visualization method showing relative characteristics of multiple documents which have different contexts. We provide users with an interactive graph correlating common words in multiple documents with characteristic words in each document, using co-occurrence and context similarity information. By allowing the users to discover underlying differences between superficially similar documents and to become conscious of them as new viewpoints, we intend to support understanding in terms of cooperative group activities or individual information collections. We examine our method using a document set with context information.