Coordinated multiple views: A critical view
Hundreds of papers have been presented at the five conferences on coordinated and multiple views (CMV) in exploratory visualization and related conferences on information visualization and visual analytics. However, CMV are still rarely implemented in commercial systems and many users are even not aware that multiple views are useful and coordination can help them to solve their problems. Even toolkits especially designed for information visualization often do not offer any coordination mechanisms for multiple views. One of possible reasons for this may be that existing CMV tools and approaches are insufficiently suited to real-life problems. Most of the CMV tools deal with single-table data sets of rather small sizes. A typical implementation allows one to create several displays (statistical graphics and/or geographical maps) representing individual entities and/or aggregates (e.g., histograms) and link them by brushing and selection. The displays are coordinated through exchanging references to selected entities. To our best knowledge, this approach works efficiently for tables with sizes up to about 105 records. Linking displays that reflect larger tables causes significant delays.