Graphical summary of result set of search, with details on demand

Map keywords into user-specified concepts

Concept can be computed as union (OR) or intersection (AND) of its constituent keywords

Automatic clustering of documents, based on their concept profile

Cluster icons arranged on sphere or 2D surface; better clusters near the "North Pole"

Relationship among clusters illustrated by color-coded arcs representing set-difference

User can open a cluster to view constituent document titles, arranged according to title similarity and score assigned by search engine

User can mark documents and clusters as good / bad / unsure and apply display filter to result set

User can see document text with color-coded keywords, via HTML

User can see textual summary of all current clusters via HTML