A data storage and retrieval system with optimized categorization of
information items using intelligent category selection, based on the
information items that are to be checked for category membership. The set
of categories for which an information item may be a candidate member is
determined, and only that set of categories is checked for membership of
the information item during the categorization process. The determination
of which categories an information item can be a candidate member of is
based on the properties defined for that information item, which are
compared to the properties checked for category membership of specific
categories. A category index tree may be built and maintained to support
efficient determination of which categories a given information item may
be a candidate member of. The efficiency of categorization provided by
the disclosed techniques enables the disclosed system to maintain
category definitions indefinitely with minimal performance impact.