Query Containment for Conjunctive Queries With

The management of semistructured data has recently received signiicant attention because of the need of several applications to model and query large volumes of irregular data. This paper considers the problem of query containment for a query language over semistructured data, StruQL0, that contains the essential feature common to all such languages, namely the ability to specify regular path expressions over the data. We show here that containment of StruQL0 queries is decidable. First, we give a semantic criterion for StruQL0 query containment: we show that it suuces to check containment on only nitely many canonical databases. Second , we give a syntactic criteria for query containment, based on a notion of query mappings, which extends containment mappings for conjunctive queries. Third, we consider a certain fragment of StruQL0, obtained by imposing restrictions on the regular path expressions, and show that query containment for this fragment of StruQL0 is NP complete. 1 Introduction The management of semistructured data has recently received signiicant attention because of the need of several applications to model and query large volumes of irregular data 1, 5]. For example, researchers in biology store their data in structured les in various data exchange formats. Similarly, large volumes of online documentation, document collections and program libraries are available in structured les. Several characteristics distinguish semistructured data from relational and object-oriented data. Unlike traditional data that ts a pre-existing and xed schema, semistructured data is irregular: attributes may be missing , the type and cardinality of an attribute may not be known or may vary from object to another, and the set of attributes may not be known in advance. Furthermore , the schema of semistructured data, even if it