Querying Semistructured Data Based on Schema Matching

Traditional database management requires design and ensures declarativity. In the context of semistructured data a more flexible approach is appropriate due to missing schema information. In this paper we present a query language based on schema matching. Intuitively, a query is a pair consisting of what we want and how we want it. We propose that the former can be achieved by matching a (partial) schema and the latter by specifying additional operations. We describe in some detail our notion of schema covering various concepts typically found in query languages, such as predicates, variables and paths. We outline the optimization potential that this modular approach offers and discuss how we use constraints for query processing.