Towards Aggregated Answers for Semistructured Data

Abstract

Semistructured data [5],[34],[23],[31],[1] are used to model data transferred on the Web for applications such as e-commerce [18], biomolecular biology [8], document management [2],[21], linguistics [32], thesauri and ontologies [17]. They are formalized as trees or more generally as (multi-)graphs [23],[1]. Query languages for semistructured data have been proposed [6],[11],[1],[4],[10] that, like SQL, can be seen as involving a number of variables [35], but, in contrast to SQL, give rise to arrange the variables in trees or graphs reflecting the structure of the semi- structured data to be retrieved. Leaving aside the “construct” parts of queries, answers can be formalized as mappings represented as tuples, hence called an- swer tuples, that assign database nodes to query variables. These answer tuples underly the semistructured data delivered as answers.