Towards Introduction of Semantics into the Infrastructure of Virtual Observatories

The objectives of the work reported consist in elaboration of the subject mediation approach to provide for the problem domain conceptual definition (including well-defined conceptions, models, theories). This definition is intended for formulation and solving of a class of scientific problems in terms of the concepts of such domain (expressed by the respective ontologies), structure and behaviour of related information entities, services and processes defined in a declarative way. Such conceptual definition of the problem domain having well-defined application and information model semantics constitutes specification of a particular mediator. A set of mediators forms a middleware located between problem specifications formulated in terms of the respective mediators and heterogeneous information resources (databases, services, ontologies, workflows). Mediation approach plays key role in semantic identification of resources relevant to a specific mediator and in their semantic integration in the specification of the problem domain (specification of a mediator). The infrastructure supporting the mediation middleware for problem solving over semantically integrated information resources has been investigated and its prototype has been implemented as a hybrid architecture integrating the AstroGrid (UK) and the facilities for the mediation middleware support (Russia). These tools are intended for problem solving in the Russian Virtual Observatory. First trials of the tools include such areas as search for distant galaxies, classification of eclipsing binaries applying data mining, interstellar extinction modeling.

This report is focused on the application of the proposed mediation approach for introducing semantics into the process of problem solving. We plan to demonstrate the mediator canonical information modeling language capabilities for conceptual definition of the interstellar extinction (ISE) domain as an example. Specific emphasis in this domain will be given to the problem of three dimensional ISE model determination over catalogs of various multicolour surveys. Specification of the problem domain semantics in a subject mediator oriented on a class of problems related to ISE includes: 1) an ontology of basic concepts of the ISE domain, 2) expressions and constants corresponding to the ISE law for the unified photometric system chosen (the Johnson - Morgan system), 3) calibrating tables for the stars of different luminosity classes, 4) the cosecant law expression for ISE, 5) specifications of types and classes providing for virtual representation of integrated data from various surveys after the data transformation to the unified photometry in different wavelength bands, 6) functions intended for cross-correlation of sources in the catalogs of various surveys. The denotations of all quantities specified in the mediator are annotated with the respective concepts of the mediator ontology. The mediator specification contains also the nonfunctional requirements to the catalogs of various surveys that should be satisfied for a survey to be applicable in our mediator. An example of one of such requirements is the existence of expressions for specific photometry system of a survey transforming it into the unified photometry system of the mediator. It is assumed also that for identification of their relevance to the mediator as well as for their semantic integration in the mediator all catalog specifications are represented in terms of types, classes and functions annotated by respective ontologies. Specification of the problem of three dimensional ISE model determination is given in terms of the mediator to be implemented in the mediation architecture.