Ed, Kingsley, and Michael, (01)
I mostly agree with Ed's note, but I'd like to add some further
qualifications to the following: (02)
KI
>> RDF does enable the incorporation of FOL into structured data
>> representation. (03)
EJB
> NO. It doesn't. RDF has a restricted FOL interpretation... (04)
Yes, but RDF can be used as input to any query and inference system. (05)
EJB
> But every RDF predicate you invent has whatever interpretation
> the author gives it, and RDF does not give the author a way
> to formally define a predicate. (06)
I agree. But that was Guha's intent. He had been the associate
director of Cyc, and he recognized the difficulties that most
programmers face in trying to learn the CycL version of logic. (07)
Guha wanted a very simple logic that could be used as a painless
intro to logic. I sympathize with that goal. But I wish that
he had included Datalog (even with angle brackets) as a language
for specifying definitions, constraints, and queries. (08)
Bob Kowalski taught Datalog to children. Even adults whose minds
were warped by SQL or SPARQL can easily learn and use Datalog. (09)
KI
> Yes, but times have changed. (010)
No. Everything you mention was implemented in the 1960s and
early '70s. Much more was foreseen -- and published -- by
people like Alan Turing and Vannevar Bush. (011)
KI
> The Internet and Web provide infrastructure that didn't exist... (012)
From http://en.wikipedia.org/wiki/ARPANET
> The earliest ideas for a computer network intended to allow general
> communications among computer users were formulated by computer scientist
> J. C. R. Licklider of Bolt, Beranek and Newman (BBN), in April 1963,
> in memoranda discussing his concept for an "Intergalactic Computer Network".
> Those ideas contained almost everything that composes the contemporary
>Internet. (013)
URLs are based on the same conventions as the Unix file system with the
identifier of the computer that supports each file system at the top. (014)
GML was invented at IBM in 1969, standardized as SGML, and adopted by
the WWW project at CERN. But the hypertext ideas were proposed by
Ted Nelson in 1963 with influence from V. Bush's article of 1945.
Doug Englebart independently developed related ideas in 1962, which
he presented in 1968 in "The Mother of All Demos". See (015)
http://en.wikipedia.org/wiki/The_Mother_of_All_Demos (016)
KI
> when deductive databases, OODBMS, and ORDBMS products took their first
> cracks at alternatives to SQL RDBMS products (017)
The first object-oriented programming language was Simula-67, which was
implemented in 1967. Simula was, and still is, an excellent OO language.
It influenced *every* OO language and system, and it is a much better
implementation of the OO paradigm than C++. (018)
The first DB systems were network (GE's IDS from 1963) and hierarchical
(IBM's IMS from 1966). During the 1970s, there were "database wars"
between the proponents of network DBs (led by Charlie Bachman, who
had designed IDS) and relational DBs (led by Ted Codd and Chris Date). (019)
Deductive databases were nearly all implemented on top of RDBs,
not network DBs. Most academics flocked to RDBs because of their
foundation in logic. At UC Berkeley, Stonebraker and his group
Ingres and QUEL, which became very popular at universities. (020)
RDBs won the day when Chris Date showed how two pages of code for the
"navigational" methods of network DBs could be replaced by a few lines
of SQL. Bachman tried to defend the network model by claiming that
"programmers enjoy a challenge". But that line killed network DBs. (021)
Bachman, by the way, invented Bachman diagrams, which Peter P. Chen
adopted with minor modifications for E-R diagrams. But those diagrams
can be used just as well to specify a database of any kind. (022)
KI
> We can teach SPARQL like SQL and have great results. I've already
> performed this experiment with students ranging from 6 - 22... (023)
MB
> My experience tells me that SPARQL queries get complicated quickly
> - especially if disjunctions are involved - so I think [John is] right. (024)
I agree with Michael. All the notations are similar with conjunctions,
SQL is simpler than SPARQL for complex queries, but Datalog is best. (025)
MB
> I wonder if Datalog is really more powerful than OWL DL as both are
> decidable ? (026)
OWL DL is restricted to tree-structured models, because that's the basis
for their proofs of decidability. You can say that a benzene molecule
has 6 carbon atoms, but you can't say that they're connected in a ring.
In Datalog, you can specify the ring. (027)
MB
> Regarding your undecidability mantra, why don't you go for OWL Full
> instead, which seems to be undecidable ? (028)
I have no objection to using DLs for specifying the T-box. In fact, I
would be happy with an even simpler T-box for most purposes. But you
can't implement a complete application in OWL. You can in Prolog. (029)
MB
> Do you know the quote from timbl in this Wikipedia article:
>
> http://en.wikipedia.org/wiki/Rule_of_least_power (030)
I came across it. Since the date is 1998, he may have been influenced
by Guha, who designed RDF as a simpler system than Cyc. But Guha
agreed with Lenat that decidability is not the issue. (031)
In any case, expressive power and simplicity are unrelated. If you
start with RDF (including blank nodes) and add negation, that is a
very simple notation. But it supports full FOL. See (032)
http://www.jfsowa.com/pubs/egtut.pdf (033)
All the restrictions in OWL DL make the language much harder to learn
than a simple version of FOL, such as Peirce's EGs. (034)
John (035)
_________________________________________________________________
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 (036)
|