Abstract This paper presents a novel approach to automatic management of large collections of electronic documents. Our approach allows a user to specify a consistent set of rules for characterizing interesting or relevant document classes. These rules, in conjunction with existing and descriptive document metadata, are used for computing all classes a document belongs to. Not only does our technique automatically populate existing document classes but it also creates new classes and deletes existing ones when new documents are added or existing ones are deleted. Furthermore, classification rules can be added and deleted at runtime, resulting in flexible classification schemes that can be easily adapted to new or changing requirements. The major technical contribution of our approach is the seamless integration of features found in existing systems (file systems, document repositories and directories, classification hierarchies, and email folders) into a single folder-based system. The resulting system is easy to use and offers great flexibility.

Top