Ontologies and Functional Dependencies for Data Integration and Reconciliation

Integrating data sources is the key success of business intelligence systems. The exponential growth of autonomous data sources over the Internet and enterprise intranets makes the development of integration solutions more complex. This is due to two main factors: (i) the management of the source heterogeneity and (ii) the reconciliation of query results. To deal with the first factor, several research efforts proposed the use of ontologies to explicit semantic of each source. Two main trends are used to reconcile the query results: (i) the supposition that different entities of sources representing the same concept have the same key a strong hypothesis that violates the autonomy of sources. (ii) The use of statistical methods which are riot usually suitable for sensitive-applications. In this paper, we propose a methodology integrating sources referencing shared domain ontology enriched with functional dependencies (FD) in a mediation architecture. The presence of FD gives more autonomy of sources in choosing their primary keys and facilitates the result reconciliation. Our methodology is validated using dataset of Lehigh University Benchmark.