Rectangular Decomposition Heuristics for Documentary Databases

Abstract This paper proposes a new algorithm for rectangular decompositon. Rectangular decomposition allows a classification of databases presented as a binary relation; it has many applications in computer science, such as data knowledge engineering, signature files, and database clustering. Existing methods for rectangular decomposition are limited to restricted relation sizes. A new algorithm for incremental rectangular decomposition is developed for a documentary database, and its complexity and cost analysis are presented.