Data integration is the problem of combining data residing at different sources, and providing the user with a unified view of these data. The problem of designing data integration systems is important in current real world applications, and is characterized by a number of issues that are interesting from a theoretical point of view. This document presents on overview of the material to be presented in a tutorial on data integration. The tutorial is focused on some of the theoretical issues that are relevant for data integration. Special attention will be devoted to the following aspects: modeling a data integration application, processing queries in data integration, dealing with inconsistent data sources, and reasoning on queries.

We show how to interoperate, semantically and inferentially, between the leading Semantic Web approaches to rules (RuleML Logic Programs) and ontologies (OWL/DAML+OIL Description Logic) via analyzing their expressive intersection. To do so, we define a new intermediate knowledge representation (KR) contained within this intersection: Description Logic Programs (DLP), and the closely related Description Horn Logic (DHL) which is an expressive fragment of first-order logic (FOL). DLP provides a significant degree of expressiveness, substantially greater than the RDFSchema fragment of Description Logic.

Although the OWLWeb Ontology Language adds considerable expressive power to the Semantic Web it does have expressive limitations, particularly with respect to what can be said about properties. We present ORL (OWL Rules Language), a Horn clause rules extension to OWL that overcomes many of these limitations. ORL extends OWL in a syntactically and semantically coherent manner: the basic syntax for ORL rules is an extension of the abstract syntax for OWL DL and OWL Lite; ORL rules are given formal meaning via an extension of the OWL DL model-theoretic semantics; ORL rules are given an XML syntax based on the OWL XML presentation syntax; and a mapping from ORL rules to RDF graphs is given based on the OWL RDF/XML exchange syntax. We discuss the expressive power of ORL, showing that the ontology consistency problem is undecidable, provide several examples of ORL usage, and discuss how reasoning support for ORL might be provided.

Abstract. Many organizations nowadays face the problem of accessing existing data sources by means of flexible mechanisms that are both powerful and efficient. Ontologies are widely considered as a suitable formal tool for sophisticated data access. The ontology expresses the domain of interest of the information system at a high level of abstraction, and the relationship between data at the sources and instances of concepts and roles in the ontology is expressed by means of mappings. In this paper we present a solution to the problem of designing effective systems for ontology-based data access. Our solution is based on three main ingredients. First, we present a new ontology language, based on Description Logics, that is particularly suited to reason with large amounts of instances. The second ingredient is a novel mapping language that is able to deal with the so-called impedance mismatch problem, i.e., the problem arising from the difference between the basic elements managed by the sources, namely data, and the elements managed by the ontology, namely objects. The third ingredient is the query answering method, that combines reasoning at the level of the ontology with specific mechanisms for both taking into account the mappings and efficiently accessing the data at the sources.

Knowledge acquisition is a difficult, error-prone, and time-consuming task. The task of automatically improving an existing knowledge base using learning methods is addressed by the class of systems performing theory refinement. This paper presents a system, Forte (First-Order Revision of Theories from Examples), which refines first-order Horn-clause theories by integrating a variety of different revision techniques into a coherent whole. Forte uses these techniques within a hill-climbing framework, guided by a global heuristic. It identifies possible errors in the theory and calls on a library of operators to develop possible revisions. The best revision is implemented, and the process repeats until no further revisions are possible. Operators are drawn from a variety of sources, including propositional theory refinement, first-order induction, and inverse resolution. Forte is demonstrated in several domains, including logic programming and qualitative modelling.

Data integratio n systemspro vide accessto a seto fhetero - geneo us, auto no mo us data so urces thro ugh a so -called glo bal schema. There are basically two appro aches fo r designing a data integratio n system. In the glo bal-centric appro ach,o ne defines the elementso f the glo bal schema as viewso ver the so urces, whereas in the lo cal-centric appro ach, o e characterizes the so rces as viewso ver theglo al schema. It is well kno wn that pro cessing queries in the latter appro ach is similar to query answering with inc o plete infoC atio , and, therefo9 is a c o plex task. On theo ther hand, it is a co mmo no pinio n that query pro cessing is much easier in the fo rmer appro ach. In this paper we sho w the surprising result that, when theglo al schema is expressed in the relatio al mo del with integrity c o straints, eveno f simple types, the pr o lemo f inco6 plete info rmatio n implicitly arises, making querypro cessing di#cult in the glo al-centric approC h as well. We thenfo cuso n glo al schemas with key andfo eign key co straints, which represents a situat io which is veryco#=W in practice, and we illustrate techniques fo e#ectively answering queries po sed to the data integratio n system in this case. 1

Partial deduction and driving are two methods used for program specialization in logic and functional languages, respectively. We argue that both techniques achieve essentially the same transformational effect by unification-based information propagation. We show their equivalence by analyzing the definition and construction principles underlying partial deduction and driving, and by giving a translation from a functional language to a definite logic language preserving certain properties. We discuss residual program generation, termination issues, and related other techniques developed for program specialization in logic and functional languages.

We present a bottom-up operational procedure for computing well-founded models of allowed DATALOG programs with negation. This procedure provides a practical method of handling programs that involve unstratified negation in a manner that may be mixed with other evaluation approaches, such as semi-naive evaluation. We also define classes of programs and sips for which the magic sets transformation preserves well-founded models with respect to the query. The class of programs and sips we consider strictly subsume those already considered in the literature, and include stratified programs (with any choice of sips), modularly stratified programs (with left-to-right sips) and programs with three-valued well-founded models (with well-founded sips). For these programs and sips, our procedure for computing well-founded models is applicable to the magic programs, thus allowing increased efficiency by specializing a program for a query. 1 Introduction Much research has been done in recent year...

The problem of modeling semi-structured data is important in many application areas such as multimedia data management, biological databases, digital libraries, and data integration. Graph schemas (Buneman et al. 1997) have been proposed recently as a simple and elegant formalism for representing semistructured data. In this model, schemas are represented as graphs whose edges are labeled with unary formulae of a theory, and the notions of conformance of a database to a schema and of subsumption between two schemas are defined in terms of a simulation relation. Several authors have stressed the need of extending graph schemas with various types of constraints, such as edge existence and constraints on the number of outgoing edges. In this paper we analyze the appropriateness of various knowledge representation formalisms for representing and reasoning about graph schemas extended with constraints. We argue that neither First Order Logic, nor Logic Programming nor Frame-based languages are satisfactory for this purpose, and present a solution based on very expressive Description Logics. We provide techniques and complexity analysis for the problem of deciding schema subsumption and conformance in various interesting cases, that differ by the expressive power in the specification of constraints.

Abstract. There are basically two approaches for designing a data integration system. In the global-as-view (GAV) approach, one maps the concepts in the global schema to views over the sources, whereas in the local-as-view (LAV) approach, one maps the sources into views over the global schema. The goal of this paper is to relate the two approaches with respect to their expressive power. The analysis is carried out in a relational database setting, where both the queries on the global schema, and the views in the mapping are conjunctive queries. We introduce the notion of query-preserving transformation, and query-reducibility between data integration systems, and we show that, when no integrity constraints are allowed in global schema, the LAV and the GAV approaches are incomparable. We then consider the addition of integrity constraints in the global schema, and present techniques for query-preserving transformations in both directions. Finally, we show that our results imply that we can always transform any system following the GLAV approach (a generalization of both LAV and GAV) into a query-preserving GAV system. 1