Dynamic Hierarchical Clustering

We describe a new method for dynamically clustering hierarchical data which maintains good clustering in the presence of insertions and deletions. This method, which we call Enc, encodes the insertion order of children with respect to their parents and concatenates the insertion numbers to form a compact key for the data. We compare Enc with some more traditional approaches and show in what circumstances Enc is eeective. Our analysis is based on simulations using queries derived from the OO7 benchmark. Our results show that our dynamic hierarchical storage method is very eecient for hierarchical queries and performs reasonably well for random access queries. Thus, using our method, hierarchical relationships between objects can be better supported in relational databases and in object-oriented databases.