Implementation and evaluation of a browsing algorithm for design applications

The implementation and evaluation of an extended relational database called ADAM (advanced database abstraction mechanism) are discussed. A browsing algorithm is developed for composite objects to support reuse of the objects. The algorithm takes advantage of aggregation hierarchies to select related composite objects and to order them based on the two measures, i.e. the relative difference of matching components (RDMC) and the number of nonmatching components (NNC). The algorithm has been implemented and various kinds of experiments have been carried out with a standard integrity constraint database. The results show that the CPU time roughly linearly depends on the number of selected objects. Around 90% of the CPU time is consumed in calculating RDMC and NNC. RDMC is more sensitive to retrieval conditions than NNC