Panel on
                  Next Generation Database Systems 
                   Won't Work Without Semantics!

                              John Mylopoulos
                          University of Toronto
                             Toronto, Canada  
                               416-978-5180
                            jm@cs.toronto.edu


 ABSTRACT

In the late '70s, while second generation DBMS products and technologies were
entering the market,  there was significant research activity whose aim was to
make greater use of semantic information  in database systems. The focus of
that research was primarily on semantic data models and data modelling,
including semantic query processing and integrity checking. However, few if
any of the results of these efforts found their way in database technologies
of the day, or database management practices. Instead, semantic issues were
delegated to early phases of information system development (including
requirements analysis and design), as well as application development.

Today's database system technologies perform admirably well with semantically
trivial operations and representations.  At the same time, these technologies
are being challenged in virtually every area of data management, with new
applications which demand ways of dealing more explicitly with the meaning of
the data being managed. For example, interoperation between applications
requires that the  underlying databases interoperate meaningfully. This
currently requires a mammoth manual reverse engineering effort that simply
cannot be sustained or funded by any large organization. The same applies to
data warehouses, since  they too require the correct semantic merging of data
from semantically diverse sources. Errors in merging these sources can lead to
significant problems of interpretation and potentially of the functions that
the warehouse is designed to deliver. The effectiveness of database
technologies to web-related information gathering and management applications
is likewise limited by the degree to which they can accommodate semantics of
the information being sought. Along the same lines, the emergence of
organizational knowledge management as the next major  application of
computing in organizations clearly offers a tremendous opportunity for
database technologies. But, again,  this opportunity begs the question whether
such technologies can succeed if they continue to ignore semantic issues.

In summary, semantic issues were put aside by database technologies of the
past. However, the database application challenges of the T90s seem to demand
solutions to precisely such issues today. This panel intends to examine these
long standing  research issues on database semantics  and their failures to
penetrate database technologies. The discussion will also review emerging
application areas and their need for mechanisms that deal with data
semantics. Finally, the panelists will comment on relevant research tasks that
need to be  addressed in this long-ignored area of database modeling,
management,  access, and processing.

Panel Questions

-- What were the major attempts to enhance data management with semantic
representations and processing and why did they fail ?
-- What current areas of data management, access, and processing might benefit
from more expressive semantic representations and processing capabilities?
-- What are some of the research tasks that need to be addressed?