XML and Graphs for Modeling, Integration and Interoperability:a CMS Perspective

This thesis reports on a designer's Ph.D. project called “XML and Graphs for Modeling, Integration and Interoperability: a CMS perspective”. The project has been performed at CERN, the European laboratory for particle physics, in collaboration with the Eindhoven University of Technology and the University of the West of England in Bristol. CMS (Compact Muon Solenoid) is a next-generation high energy physics experiment at CERN, which will start running in 2007. The complexity of such a detector used in the experiment and the autonomous groups that are part of the CMS experiment, result in disparate data sources (different in format, type and structure). Users need to access and exchange data located in multiple heterogeneous sources in a domain-specific manner and may want to access a simple unit of information without having to understand details of the underlying schema. Users want to access the same information from several different heterogeneous sources. It is neither desirable nor feasible to require users to have knowledge of these schemas and different storage formats. Instead it would be advantageous if a user could query or exchange data using his/her own domain models and abstractions of the data in a predefined data format. This thesis identifies the different integration problems and approaches within CERN and CMS. The thesis discusses a generic structure of directed acyclic graphs that can be found in many CMS and CERN databases. This generic structure is the basis of the detector description project described in this thesis, and is used to describe a generic architecture for specification of mediators for CMS data sources.