Free University of Bozen-Bolzano, in the main building – room D102 (first floor).

• Opening
• Student presentations
  • Luciana Benotti
    title: Enhancing a Dialogue System Through Dynamic Planning
  • Lina Lubyte
    title: Extracting and Materializing the Conceptual Schema from a Relational Database
    abstract: A number of important database problems have been shown to have improved solutions by using a conceptual model (an ontology) to provide precise semantics for a database schema. At its best, such semantics of the data is captured by some kind of semantic mapping between the database schema and the conceptual model. In this talk we will concentrate on the problem of extracting and materializing the conceptual schema of a relational database designed according to Entity-Relationship techniques. To uncover the connections between the schema and a formal conceptual model, at the first part of the talk we will revisit standard database design principles from ER diagrams, thus obtaining ER-to-Relational methodology that defines correct relational representations from ER model to relational model. We fully exploit the constraints on ER objects in order to capture the information, including primary and foreign keys, uniqueness constraints, null values, domain constraints, etc. At the second step, we devise the process for extracting ER model constructs together with a set of views, assuming to have the input database designed with our proposed ER-to-Relational methodology. The semantic mapping between the database schema and the conceptual model is thus expressed, where every view, defined over the actual data, corresponds to a unique ER construct.
  • Luis Angel Montiel Moreno
    title: Safe Beliefs Framework
    abstract: Framework Safe Beliefs. A generalization of the notion of answer sets for arbitrary propositional theories. Reductions, equivalence between programs, presentation of an algorithm, based on the Davis-Putnam method, to compute safe beliefs for arbitrary propositional theories.
  • Evgeny Kharlamov
    title: Model Theory and Calculus for The Description Logic DL-Lite
    abstract: We investigate two aspects of DL-Lite: properties of models of DL-Lite knowledgebases (KBs) and a calculus for the logic. We construct examples to illustrate that in a general case DL-Lite KBs have neither finite nor least (wrt inclusion) models. We introduce the notion of a universal model for a KB and show that any satisfiable DL-Lite KB has a universal model. We also show that a chase of a knowledge base is a universal model. We show that for answering conjunctive queries (CQs) over a DL-Lite KB it suffices to evaluate it only over a universal model of the KB, and consequently, over a chase of the KB. In general the chases of a KB may be infinite. We identify a class of KBs for which all chases are finite and, moreover, CQs can be answered in polynomial time. However, it turns out that, also for KBs that have an infinite chase, it is possible to answer CQs in finite time. We do this by defining a calculus that takes as input a KB and a CQ and allows one to derive all answers. We show how a rewriting algorithm presented by Calvanese at all. can be obtained by imposing a specific control strategy on the calculus.

• Diego Calvanese
• Raffaella Bernardi
• Sergio Tessaris

• Student presentations
  • Szilvia Halasz
    title: Optimizing Perfect Rewriting in DL-Lite
  • Magdalena Ortiz
    title: Data Complexity of Query Answering in Expressive Description Logics
    abstract: DLs/ontologies are increasingly seen as a mechanism to query data repositories. This calls for investigating flexible query mechanisms over expressive knowledge bases. In particular, very expressive description logics must be considered (in oder to capture common data modeling constructs) in combination with well established query languages, such as those inspired by database technology. Moreover, since the extensional data (ABox) is usually much larger thanthe terminological component (TBox), it is crucial to study data complexity, i.e. to single out the contribution of the ABox in the overall complexity of reasoning, in order to optimize the inference techniques with respect to the data size. In this talk, we will present a tableaux-based algorithm for checking entailment of conjunctive queries and unions of conjunctive queries in the DLs SHIQ and SHOIQ. The standard blocking conditions of the SHIQ/SHOIQ tableau are replaced by some modified ones, which make it suitable for answering all queries that do not contain transitive roles. The algorithm provides a coNP upper bound in data complexity. This, together with a long known matching lower bound for logics even weaker than ALC, shows that a wide range of DLs are coNP complete in data complexity. Some other applications of the novel tableaux-based algorithm are discussed. Notably, since the algorithm is inspired by work on hybrid knowledge bases, it allows us to extend the CARIN family of languages to CARIN-SHOIQ, a hybrid knowledge representation language combining very expressive DLs and non-recursive Datalog rules, and provide a sound and complete reasoning algorithm for it.
• Question Time
• Handing of Diploma Supplement and Best Thesis award