The problem of integrating multiple information sources into a uni?ed data store is currently
one of the most important challenges in data management. Within the ?eld of source integration
the problem of automatically gen- ating an integrated description of the data sources is surely
one of the most relevant. The signi?cance of the issue can be best understood if one c- siders
the huge number of information sources that an organization has to integrate. Indeed it is
even impossible to try to do all the work by hand. Like other important issues in data
management the problem of integrating multiple data sources into a unique global system has
several facets each of which represents per se an interesting research problem and
comprises for instance that of recognizing at the intensional level similarities and
dissimilarities among scheme objects that of resolving representation m- matches among schemes
and that of deciding how to obtain an integrated data store out of a set of input sources and
of a semantic description of their contents. The research and application relevance of such
issues has attracted wide interest in the database community in recent years. And as a con-
quence several techniques have been presented in the literature attacking one side or another
of this complex and multifarious problem.
