Related Communities:

UNIMOD Project

RFBR Grant 06-07-08072

Development and prototyping of experimental system for constructing the unifying information representation models for interoperable integrating systems of heterogeneous information sources (UNIMOD)

Project starting date

July 2006

Project duration

21 months

Project co-ordinator

Igor Sokolov, Institute of Informatics Problems RAS, Moscow, Russia

Participants

  • Institute of Informatics Problems RAS

Objectives

The present period of IT development is characterized by the process of explosive growth of various information representation models. This development takes place in frame of specific distributed infrastructures (such as OMG architectures (in particular, the model driven architecture (MDA)), SemanticWeb and Web service architectures, digital library architectures as collective memories of information in various subject domains, architectures of the information Grid), as well as in the standards of languages and data models (such as, for example, ODMG, SQL, UML, XML and RDF stacks of data models), process models and workflow models, semantic models (including ontological models and models of metadata), models of digital repositories of data and knowledge in particular domains.

This process is accompanied by another trend — intensive development of based on such models information components and services. This growth causes the accelerating need for integration in various applications of components and services represented in heterogeneous models, as well as their reuse and composition implementing new information systems. The more variety of applied models we meet in various components and services, the more complex become problems of their integration and composition. Research and development of adequate methods for manipulation of various information models during information systems design are required.

The basis of these methods is constituted by the concept of a canonical information model serving as the common language, ”esperanto”, for adequate uniform expression of semantics of various information models surrounding us. Intensive researches made by the SYNTHESIS group for the canonical information model synthesis are based on the following principles:

  • Canonical model extensibility principle. The canonical model should be extensible in process of new information model embedding (such extension can be axiomatic, generic types based, metaframe based, etc.) The result of the extension should be refined by the resource information model. A kernel of the canonical model is fixed.
  • Information model commutative mapping principle. In the process of mapping of an information model into the canonical one information and operations should be preserved. To prove that formal specification facilities (Abstract Machine Notation) are applied.
  • Unifying canonical model synthesis principle Canonical model synthesis is a process of construction of extensions of the canonical data model kernel refined by information models included into an environment of the information system and a process of merging of such extensions into a canonical model.

The project for developing specific instrumentation for the canonical model synthesis (Information Model Unifier) is based on the following results obtained by the SYNTHESIS group:

  • Canonical model kernel has been defined (the SYNTHESIS language)
  • The Abstract Machine Notation (AMN) has been used as a formal metamodel. It is exploited to define the model-theoretic specifications in the first order logics and to prove the fact of specification refinement
  • Mapping of the canonical model kernel into AMN (formal semantics of the kernel) has been constructed
  • Extensions of the canonical model kernel and the respective refinement proofs have been obtained for:
    • Canonical model kernel has been defined (the SYNTHESIS language)
    • The Abstract Machine Notation (AMN) has been used as a formal metamodel. It is exploited to define the model-theoretic specifications in the first order logics and to prove the fact of specification refinement
    • Mapping of the canonical model kernel into AMN (formal semantics of the kernel) has been constructed
    • Extensions of the canonical model kernel and the respective refinement proofs have been obtained for:

Taking into account the labor intensive character of development of the required information model mappings (compilers) specific Meta Environment facilities (Brand, 2002) are used for declarative specification of mappings and generating them (applying term rewriting technique) according to such specifications.

The main objective of the current research phase is a design of a prototype of a constructor of unifying information representation models for mediators and interoperable systems integrating heterogeneous information sources. On the final stage of the project a prototype of the Model Unifier is planned to be developed in accordance with its architecture that has been designed, canonical information model kernel and its ontology, a scenario of the Unifier functioning. The Unifier is intended to be used as one of the instruments for specific information systems construction as well as an intersystem facility for the Centers of accumulating knowledge about various information models applied in the world, their semantics and their unifying transformation into the canonical model.. The methods applied by the SYNTHESIS group for the Model Unifier development are distinguished from the known ones (e.g., ModelGen investigated by Microsoft Research). We emphasize the extensible canonical model constructed by means of its kernel extensions refined by specific resource models, provision of formal facilities for information model specifications and proof of refinement of their mappings into the canonical one. SYNTHESIS is not limited by structured data models, but is considered a broad class of information models. In contrast to ModelGen that is abstracted from actual information models (languages) taking from them only data structuring facilities, the Model Unifier considers canonical and resource information models as languages with all their details. Application of the Model Unifier provides complete compiler as a product for the information model mapping.

Related Publications

Supported by Synthesis Group