Joint project-team with LRI (CNRS and University of Paris-Sud) and LIX (CNRS and Ecole Polytechnique).
VERSO
Databases
Serge Abiteboul
Type :
Project-Team
VERSO was dissolved on 31 December 2002
Team presentation
Data handled by information systems is increasingly complex,
distributed, heterogeneous, replicated, changing, and multi-formatted.
The goal of the Verso project is to study fundamental problems and
develop novel techniques for the management of such data. The goal is
to build richer systems that are more open to the network and other
software components.
The integration of data from a very large number of data sources is
becoming increasingly important in the context of the Web. Such
integration often requires combining techniques from artificial
intelligence and databases. To facilitate this research, Verso has
formed in 2001 a joint project together with the knowledge
representation team (IASI) of the Laboratoire de Recherche en
Informatique at Orsay (led by M.-C. Rousset).
Research themes
As a general trend, Verso is moving away from the centralized server
approach used in Active Views and Xyleme towards peer-to-peer
solutions as exemplified by the Active XML approach. Specific research
topics include the following:
Warehousing XML data from the Web: pursuing work initiated in
the framework of Xyleme, we study algorithms to detect changes in XML
documents (XMLdiff), query optimization and the computation of page
importance on the Web. We also started a new reserach direction on
the creation and mining of thematic data collections.
Active XML: we study the integration of data and Web services.
Our approach is centered around XML documents that include declarative
calls to Web services. The goal is to facilitate the development of
applications open to the network, and the cooperative work of
distributed clients.
Querying heterogeneous data and optimization: we continue the
work on XML indexing and heterogeneous data integration. New work
deals with filtering of XML data using finite-state automata.
Theory of data modeling : we study fundamental problems
motivated by a data centric vision of computer science. We use tools
from logic and complexity theory to understand the specific properties
of computation on collections (relations) or irregular graphs (such as
the Web).
International and industrial relations
Initiated the DbGlobe European project on query processing over
the Web.
Initiated a French-Canadian project with U. Toronto on new
services for the web.
Within France, collaborations with the Vertigo Group, CNAM, Paris
Industrial collaborations with the MatchVision start-up (that
disappeared in 2001, GAEL RNRT Project) and Xyleme, a start-up
originating in Verso.