TagStar visualizes the indexing quality of classified resources.
This tool integrates two views: one for visual analysis, and one for indexing. The first view for visual analysis loads a large amount of resources from a database and displays them simultaneously. It represents each resource with a glyph that reflects the attached index terms and their relation to the underlying classification scheme. Resources that do not comply with the indexing strategy pop out and indicate items that need revising.
The second view shows only one resource in detail. It acts as an indexing interface and allows the indexer to add index terms from the controlled vocabulary or remove assigned index terms. The view uses the same glyph construction principle, and recommends index terms by visualizing their probability for particular resources.
This concept addresses personal and collaborative information systems that need the flexibility of social tagging systems, such as free terms for classification or multiple indexers. However, it also helps to sustain a consistent base for information-seeking maintained by only a few people. Since both approaches are contradictory to some extent, TagStar aims at the combination of both approaches in order to classify dynamic data such as web resources. It supports indexers when they create or adapt controlled and preferably faceted vocabularies. While rearranging facets, adding index terms to the vocabulary, or moving terms from one facet to another, the visualization could be used to evaluate the effects on the findability of resources in a faceted search scenario.