LIP6 2005/002
- Habilitation
Bases de données : de l'objet à l'interopérabilité - B. Defude
- 81 pages - 02/24/2005- document en - http://www.lip6.fr/lip6/reports/2005/lip6-2005-002.pdf - 631 Ko
- Contact : Bruno.Defude (at) nullint-evry.fr
- Ancien Thème : OASIS
- Keywords : databases, object-oriented DBMS, integrity constraints, interoperability, integration of spatial schema, metadata, semantic web, adaptation to users
- Publisher : Thierry.Lanfroy (at) nulllip6.fr
The database area has completely evolved in the last 15 years considering global software evolution (objects, components, web), hardware evolution (exponential increase in CPU) and network evolution (generalized connectivity, Internet, mobile networks, . .). At the same time "new" applications have emerged and proposed new needs to model and manipulate data.
This memorandum presents the evolution of our research in this context and is structured in two main parts. In the first part we present projects about the coupling of databases and objects technologies.
Our main contributions are the following:
- a relational-object data model allowing extensibility using abstract data types. This work has been done in a industrial context (MATRA with a DGA-DRET grant) and has been validated by a software prototype ;
- description and enforcement of integrity constraints for object-oriented DBMS. Our proposal is based on a behavioral model where constraints are defined on operations and not on data themselves. This allows a better integration with object-oriented concepts such as inheritence, encapsulation and overloading. A mechanism for constraint enforcement has been proposed for a nested transaction model. This mechanism allows us to better define the interaction between transactional concepts and constraint enforcement. This work has been validated by a software prototype developed for a grant with CNET France Télécom ;
In the second part, we describe our projects about databases and DBMSs interoperability. This problem is crucial nowadays with massive interconnexion of information systems using Internet. Our main contributions are the following:
- a methodology and associated CASE tools to define databases applications using a model driven approach. In this work we have proposed a meta-data model which generalize object-oriented and relational data models. A database application is described using this meta-data model and then may be generated for different target DBMSs. This work has been done in the Aristote project, a joint research project between Bull research centre in Grenoble and LGI-IMAG laboratory in Grenoble;
- a pivot spatial data model and algorithms for the integration of heterogeneous spatial databases. This pivot model is based on description logics which have interesting properties for schema integration such as automatic instance classification or subsomption. The integration algorithm uses these properties to define a semi- automatic integration process;
- a description model for pedagogical resources and adaptation algorithms . The description model uses a domain model (an ontology) and is based on semantic web technology. This allows to formaly define composition operators for resources. The adaptation process will choose one possible instanciation of a resource to deliver it taking into account the user profile.
The conclusion of this memorandum includes several open issues which are interesting for our work. The first is personalised information systems (how to handle user profile at the core of systems and not only at man-machine interface level). The second concerns distributed management of numerical resources (how to handle interoperation between resources described using different ontologies). The third is large scale resource management (how to integrate semantic web and P2P systems).
This memorandum presents the evolution of our research in this context and is structured in two main parts. In the first part we present projects about the coupling of databases and objects technologies.
Our main contributions are the following:
- a relational-object data model allowing extensibility using abstract data types. This work has been done in a industrial context (MATRA with a DGA-DRET grant) and has been validated by a software prototype ;
- description and enforcement of integrity constraints for object-oriented DBMS. Our proposal is based on a behavioral model where constraints are defined on operations and not on data themselves. This allows a better integration with object-oriented concepts such as inheritence, encapsulation and overloading. A mechanism for constraint enforcement has been proposed for a nested transaction model. This mechanism allows us to better define the interaction between transactional concepts and constraint enforcement. This work has been validated by a software prototype developed for a grant with CNET France Télécom ;
In the second part, we describe our projects about databases and DBMSs interoperability. This problem is crucial nowadays with massive interconnexion of information systems using Internet. Our main contributions are the following:
- a methodology and associated CASE tools to define databases applications using a model driven approach. In this work we have proposed a meta-data model which generalize object-oriented and relational data models. A database application is described using this meta-data model and then may be generated for different target DBMSs. This work has been done in the Aristote project, a joint research project between Bull research centre in Grenoble and LGI-IMAG laboratory in Grenoble;
- a pivot spatial data model and algorithms for the integration of heterogeneous spatial databases. This pivot model is based on description logics which have interesting properties for schema integration such as automatic instance classification or subsomption. The integration algorithm uses these properties to define a semi- automatic integration process;
- a description model for pedagogical resources and adaptation algorithms . The description model uses a domain model (an ontology) and is based on semantic web technology. This allows to formaly define composition operators for resources. The adaptation process will choose one possible instanciation of a resource to deliver it taking into account the user profile.
The conclusion of this memorandum includes several open issues which are interesting for our work. The first is personalised information systems (how to handle user profile at the core of systems and not only at man-machine interface level). The second concerns distributed management of numerical resources (how to handle interoperation between resources described using different ontologies). The third is large scale resource management (how to integrate semantic web and P2P systems).