A cost model for sort-domain traversal strategy in object-oriented databases

Some factors such as set-valued attributes, partial instantiation and clustering need to be considered to estimate the processing cost of an object-oriented query, In this paper, we propose a cost model which reflects the effect of such factors, The cost of a query is estimated using the number of objects participated in the instantiated paths and the number of tuples and the size of the reference relation which is an intermediate result of class traversals, The proposed cost model utilizes a small set of statistics maintained by the system, and estimates the costs more accurately.