Sunday  2002-12-08   Web of Services

Semantic Web is hard

Semantic Web is hard.

The Resource Description Framework and the Resource Description Framework Schema Specification are supposed to be the foundations of the Semantic Web, in that all other Semantic Web languages are to be layered on top of them. It turns out that such a layering cannot be achieved in a straightforward way. This paper describes the problem with the straightforward layering and lays out several alternative layering possibilities. The benefits and drawbacks of each of these possibilities are presented and analyzed.

Sunday  2002-12-08   Web of Services

From web to grid

Besides web services, Semantic Web has another technology it can leverage to gain viability: grid computing. Carl Kesselman, one of the leaders of grid computing gave a keynote speech at this year's International Semantic Web Conference titled "The Grid, Grid Services and the Semantic Web: Technologies and Opportunities". Naturally, this merger of the two technologies will be named Semantic Grid.

Wednesday  2002-11-27   Web of Services

WS = SW

Over the last few years Semantic Web has been dovetailing web services. Three examples:

• Tim Berners-Lee's keynote speech at WWW2002 conference. Current disparity between Semantic Web and web services is "a compromise".
• Semantic Web Working Symposium in 2001 published many papers on how Semantic Web helps web services.
• One of the keynote speeches at this year's International Semantic Web Conference was "Semantic Web enabled Web Services". Again many papers on the subject.

Tuesday  2002-11-26   Web of Services

Business process management overview - part 3

DARPA agent markup language or DAML is an application of RDF/XML to write ontologies. DAML-S is a DAML ontology for describing web services or modeling business processes. It aims to support:

• automatic service discovery (unlike UDDI)
• automatic service invocation (unlike WSDL but similar to BPEL4WS)
• automatic service composition and interoperation (not covered by either WSDL or UDDI)
• automatic service execution monitoring (ditto).

See also previous note parts one and two.

Tuesday  2002-11-26   Development

Objects have failed?

Richard P Gabriel provokes: "Objects have failed."

Monday  2002-11-25   Development

More intertwingularity

At the end of his essay on intertwingularity or phenomenon of deeply entangled inter-depedent messages and information, Jamie Zawinski writes

This sort of model [of utilizing intertwingularity] is not applicable merely to the domain of [email] messages; it applies equally well to any corpus which has structured, potentially-ambiguous references (or rather, representations of references.)

For example, source code.

Programmers have been making many tools to deal with volumes of source code. One simple way is to format the code in certain convention and use tools like grep or text editor's search commands. Another method is to use simple databases like TAGS or source code browsing tools like cscope to augment text editors. More tools like lxr (Linux Cross Reference) and global that generate web pages came during last several years. These tools however know only of source code text, not its semantics as parsing is done by compilers after editing code is finished.

While people on conventional operating systems deal with files, Smalltalk programmers store their source code as objects in real databases. Because these databases or images are written in Smalltalk and interact with users in Smalltalk (ie images know the semantics and inter-relationships expressed in the source code), users need not bother to disambiguate manually what a piece of text in question may be as they browse the code. Smalltalk browser has been evolving to many interesting directions like Whisker, Refactoring Browser and Star Browser.

Old Lisp Machines (like the ones Zawinski used) had similar features that make an old timer like Kent Pitman nostalgic and it is easy to see why.

Mainstream Java programmers are discovering productivity gains from using an environment that does more than text search but rather understands their code like IntelliJ IDEA. As more attention is focused on making sense of far too many artifacts of software builing, more tools like Hipikat that can find relevent information from "source code, documentation, bug reports and fixes, e-mail, and version information", will receive warm welcome.
 
permalink   
 

Sunday  2002-11-24   Software

Intertwingle

"Everything is deeply intertwingled."
-- Ted Nelson
"Everything is related to everything else."
-- Vladimir Ilyich Lenin

More than four years after Jamie Zawinski wrote about dealing with large volume of messages by aggressively exploiting their relationship to one another (expressed in their "metadata" and data), several new promising programs in the category of PIM or personal information manager are appearing. ZOË is hailed as Google for email for surfacing out all kinds of relevant information about a given email. Spaces tries to connect email, calendar, tasks and notes by providing multiple views on them. Mitch Kapor, famous for Lotus 1-2-3, is also designing Chandler as an inter-personal information manager - sharing information with other users is its primary design goal. Kapor is also known for Agenda, a pioneering PIM from 1980s.