HINT:Heterogenous Information Integration

The number of high quality datasets available on the web is steadily increasing. These data usually come in different formats - above all as relational, XML, and RDF data. Moreover, they are accessed via specific query languages such as SQL, XQuery, and SPARQL, respectively. However, applications are usually designed for a unified access to these diverse data. The integration of heterogeneous information is thus becoming an increasingly important challenge. Information integration has evolved as a particularly active area of database research and much progress has been made in recent years. However, most of the research so far has been devoted to relational data. Other data formats like XML and RDF have received far less attention. A systematic study of information integration in a heterogeneous setting (where relational, XML, and RDF data have to be handled at the same time) is missing to date. The goal of the HINT project is to initiate the research on heterogeneous information integration. We thus want to propose various schema mapping languages for heterogeneous settings, i.e., logical formalisms to express relationships between database instances of different formats. We will study fundamental properties of these schema mapping languages and investigate query answering as well as update propagation in such environments.

Nowadays, the data accessed and processed by an application is typically distributed over many data sources both in various databases and on the web. Therefore, the integration of data in order to allow for a uniform access is crucial. Consequently, information integration has evolved as an important sub-discipline in database research. In this project we have made many contributions to the following areas, which are all relevant to information integration: Schema mappings allow one to describe the relationship between data sources structured in different ways. In this project, we have explored fundamental properties of expressive schema mappings. As mentioned earlier, the integration of data typically includes web data. In this project, we have studied intensively the semantic web query language SPARQL. Our focus was on the efficient solution of various problems arising in the context of SPARQL query evaluation. The integration of data from various data sources can easily lead to inconsistencies. We have studied the problem of inconsistency handling for various kinds of integrity constraints, which are defined by schema mappings of differing expressive power.Transforming data from one formalism into the other poses particular challenges. We have studied such problems when transforming relational data (i.e., data stored in relational databases) into semantic web data. An important aspect of this transformation is to preserve relationships (such as functional dependencies) which may exist between different data items.Data on the web are often incomplete. Ontological reasoning allows one to use domain knowledge to fill some gaps in incomplete data. We have studied ontology-mediated query answering for two important ontology formalisms, namely Description Logics and existential rules. The evaluation of a fundamental class of queries (namely Conjunctive Queries, CQs) is intractable. Hence, the search for tractable fragments of CQ answering have been an active research field for several decades. We have extended syntactic characterizations of tractable classes of CQ answering to semantic characterizations.