Dynamic visual metaphors for news story abstractions

Abstract

This paper describes a dynamic document genre for on-line news, specifically designed to support end users in both easily skimming news stories as well as reading them in-depth. Through careful analysis of the needs of online news readers, and understanding the opportunities and constraints of the computational medium, we set out to define characteristics of a document management environment which uses linguistically-intensive content analysis techniques to derive document abstractions usable as `zooming' windows into and out of documents. The notion of flexible mediation between progressively richer, and more informative, layers of document content defines the core underlying characteristic of a dynamic document genre. We prototyped and deployed three examples of dynamic document viewers that characterize this new genre. The design incorporates a novel mix of: state-of-the-art linguistic analysis techniques that produces capsule document overviews, novel temporal typography techniques that allows us to dynamically change how these overviews are presented to the reader, specialized search and filter mechanisms to cull collections of relevant documents from the Web, and large screen displays to facilitate peripheral awareness of constantly changing news feeds.