Augmenting citation chain aggregation with article maps

Abstract

This paper presents Voyster, an experimental system that combines citation chain aggregation (CCA) and spatial-semantic maps to support citation search. CCA uses a three-list view to represent the citation network surrounding a ‘pearl’ of known relevant articles, whereby cited and citing articles are ranked according to number of pearl relations. As the pearl grows, this overlap score provides an effective proxy for relevance. However, when the pearl is small or multi-faceted overlap ranking provides poor discrimination. To address this problem we augment the lists with a visual map, wherein articles are organized according to their content similarity. We demonstrate how the article map can help the user to make relevant choices during the early stages of the search pro-cess and also provide useful insights into the thematic structure of the local citation network.