Algorithms for list structure condensation

Abstract

The need to detect and merge redundant information is a familiar problem in data management. This paper presents a particular definition of redundancy in list structures, along with algorithms for its removal. The process, termedcondensation, transforms a list structure into a minimally sized equivalent structure in which each cell has a unique information content. The resulting structure is often smaller than the original, and permits substructure equivalence testing to be done by trivial address identity comparison. Three algorithms are presented, two for noncyclic structures and one for cyclic. The best time result in both cases isO(mn) for ann-cell structure withm cell equivalence classes. In the case of noncyclic structures, that speed is attained under bounded workspace if a free mark bit is assumed in each cell. The cyclic algorithm is illustrated by application to themode equivalence problem ofalgol 68.