A system and method for applying intensional category representations to provide dynamic categorization of information items in a data storage and retrieval system is disclosed. The disclosed system may apply pattern recognition techniques during information storage and retrieval to provide dynamic categorization of information items. Information items are categorized as they are observed. Information items may each be associated with one or more categories. An information item may become a member of a category either initially, when it is created, or subsequently by dynamic acquisition of certain properties. Category membership may be further be acquired or lost when property values are changed, and/or as a result of changes to properties or property values of referenced or referring information items. The disclosed information storage and retrieval system operates to categorize objects as they evolve over time. Properties may be added to information items, thus changing the structure of the information item, with the result that new functionality is dynamically associated with the information items.