Incorporating New Languages into DL Systems

Papers Submitted to the Workshop

The great expansion of the communication networks have made available to the users of a huge number of heterogeneous and autonomous data repositories. However, these repositories present different structures/ organizations, query languages and data semantics, making very difficult for the users to access the data stored on them. A possible solution to lighten the problem of lack of uniformity when dealing with the available repositories consists on defining new information retrieval techniques with a strategy that focuses on information content and semantics. We propose to represent intensional descriptions of the objects in the repositories as metadata introducing in this way a semantic view over the repositories. Different kind of systems can be used to represent those semantic views. We consider that systems based on Description Logics (DL systems) are interesting for that purpose due to the following reasons [BIGB94; GBBI96]: they allow definition of semantically richer views, they are also appropriated to offer intensional answers to the users and last, reasoning mechanisms from DL systems are useful to perform query optimization and in particular semantic and caching optimization. Moreover, when defining semantic views over repositories it is necessary to include also a mapping information that relates terms of the semantic views (classes and roles in our case) with one or more repositories where the actual data are stored. The mapping descriptions play a key role in encapsulating the heterogeneity due to different formats and organization of the data in the various repositories. Existing DL systems provide languages for describing classes and roles and also for creating instance objects that represent the beliefs of the system. However, for the considered framework, managing multiple information systems, two more languages, one for defining meta terms and another one for describing the mapping information can be very useful. The goal of the first language is in general to provide extensibility and in our case we use it to describe the syntactic structure and the semantic interrelationships among the components of terms. The goal of the second language, that we are working on, is to allow defining mapping descriptions at the same level as classes and roles.