Presenting large amounts of information in a limited screen space is a significant challenge in the field of Information Visualization. With the rapid development and growing use of small handheld devices such as PDAs this issue has become more important. Many Focus+Context techniques have been developed to address it but very few of them would effectively aid visualization applications for small handheld devices. In this paper we propose a new approach for visualizing a clustered graph by adding filtering to geometric distortion. Different from most existing techniques, our approach focuses on contextual information. Beginning with the assumption that not all contextual information is necessary to visualize a user's interests, we propose an approach for filtering out some "useless" context to compensate for the small screen size. Samples of applying the proposed approach to visualization of a rooted tree and a clustered graph on a small screen are presented. The limita...