Querying Semistructured Heterogeneous Information

摘要: Semistructured data has no absolute schema fixed in advance and its structure may be irregular or incomplete. Such commonly arises sources that do not impose a rigid (such as the World-Wide Web) when is combined from several heterogeneous sources. Data models query languages designed for well structured are inappropriate such environments. Starting with “lightweight” object model adopted TSIMMIS project at Stanford, this paper we describe language repository specifically semistructured data. Our provides meaningful results cases where conventional not: some absent, does have regular structure, similar concepts represented using different types, sets present, fully known. This motivates key behind our approach, describes through series of examples (a complete semantics available an accompanying technical report [QRS+94]), basic architecture processing strategy developed.