Sean, (01)
There has been a long history of collaboration among DB and KB
researchers. There were many conferences devoted to such work.
One example is the series on Very Large Databases (VLDB), which
will hold its 35th annual conference in August in Lyon: (02)
http://vldb2009.org/ (03)
Another is the series of conferences on Data Semantics, which were
sponsored by IFIP Working Group 2.6. The last one was DS-8, which
was held in 1999: (04)
http://www.informatik.uni-trier.de/~ley/db/conf/ds/ds1999.html (05)
1999 is a significant year. First, it marked the end of the old
millennium. Many people wanted to make a clean break with the
past, and something new was starting up: the Semantic Web,
which sucked a lot of research euros and dollars out of the
DB+AI work. (06)
So 1999 was the last of the DS series. VLDB continued, but it
has a much smaller chunk of the hype spectrum than it had in
the BSW epoch (Before the Semantic Web). (07)
SB> Coming from a data integration background, I was not aware of
> the AI literature or work in this area, and one only has limited
> time to read around. Your comment "But the SemWebbers haven't
> yet noticed." points to a broader problem, which you have alluded
> to before, that the SemWebbers have a very narrow view of the subject. (08)
In the 1980s and '90s, there was a considerable amount of interest
in combining AI and DB technologies and methodologies. Cyc, for
example, uses a relational DB for storing their facts. They've
added interfaces to RDF and OWL, but SQL is much richer and more
convenient for integration with AI technology. (09)
If you need to make large numbers of complex queries against RDF
data, the best approach is to map the data to an OODB that also
supports SQL. (010)
Objectivity, for example, has always supported both path-based
queries and SQL. The overwhelming majority of programmers prefer
to use SQL -- so Objectivity implements the obvious optimizations
that make SQL efficient on networks. SPARQL advocates claim that
SQL queries are inefficient against web data. But that is only
because their systems aren't optimized. (011)
SB> The fact that the ideas already exist doesn't negate the need
> for a paradigm shift among SemWebbers - it just means it should
> be a lot easier. (012)
Unfortunately, most of them think they have *already* made the
paradigm shift. You can tell them that the technology for
triple stores, description logics, *ML markup languages, and
Internet protocols came from the late 1960s and early '70s.
But since none of that history is accessible via the WWW,
they don't believe it exists. (013)
The SemWebbers will not listen to any arguments for paradigm
shifts *away from* their dogma. The only strategy that might
work with them is to propose an abstraction layer that supports
SemWeb interfaces and "legacy" interfaces in a seamless integration. (014)
Under the abstraction layer, any kind of DB or KB technology with
any level of optimization could be used. The abstraction level
should support very good human interfaces. It should also
support any kind of underlying technology that was designed
for legacy systems or the current Semantic Web and any new
technologies that might be invented. (015)
Suggestion: We should propose an abstraction layer that will
simplify and optimize the next generation of the Semantic Web,
support languages with much better human factors, and support
seamless integration with legacy systems. (016)
John (017)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (018)
|