Wednesday, May 18, 2005
concept of standardizing unique definitions for words
communication to http://groups.yahoo.com/group/jena-dev/
Diego and Chris
Chris, I know that the "Jena community" is diverse. You seem defensive about the notion that there is a community, is this because you do not recognize that an alternative to the viewpoints expressed by HP and W3C are legitimate? I am just curious.
What I mean by the phrase "Jena community" is the community that is interested in the process organized by HP and W3C informally. I am a member of this community but not at the core of the community, as you are (right?). There is diversity within those who try to develop systems using Jena, but there are also well defined alternative viewpoints. Yes?
Diego, you mention that you are after performance. Several others express frustration with not being able to see the concepts in an ontology without long URI strings. Visualization is rarely mentioned.
Visualization is not occurring in a developer's community that uses Jena, in spite of the expectations that Chris and others have stated.
The URL given by Chris,
http://www.hpl.hp.com/techreports/2003/HPL-2003-266.html
mentions a performance optimization in JENA2 over what was available in JENA1, but the details are in a technical report that I have not read. In any case, the many current questions about JENA functionality and performance is about JENA2.
RDF does not have to use URIs!!!!!
If the concept definitions are all locally provided then the persistence of an RDF store becomes far easier than is now the case. The is a fact. The notion of a interconnected web is important to the W3C vision, but the fact is that concept definition can be usefully localized and then reconciliation of terminological differences handled by web services. Differences of opinion are not handled well if a large percentage of individual concepts are remotely located and requiring of long URI strings. Dublin core is an excellent example of a solid standardized set of concepts... but this standard can be applied remotely without URIs making the human reading of RDF nearly impossible. Namespaces help, as does the use of macros to shorten a namespace to a short string. But the standard is cumbersome and this cumbersome-ness is reflected in the frustrations that software engineers have with RDF (and OWL). There are very few users of OWL or RDF that are not developers trying to create something that works.
One can look at Intellidimension's RDF repository to see a highly efficient data encoding where one can achieve what appears to be far better access benchmarks than JENA2. So there are better models.
As you comment, Diego, the technical background (about JAVA and JENA specific things) is so extensive that comparisons are difficult. Intellidimension's system does not require the same level of commitment, as does JENA2, in my opinion.
And in spite of the claims by the W3C, RDF layer cake is not necessarily the future foundation of a semantic web. We may find that John Sowa's comments about the layer cake are more relevant than what W3C wants to admit today. We may find my work, or the work by Richard Ballard on n-ary representation of concepts more valid than the RDF triple. And we may find the fixation on standard logics simply misplaced.
Jena is a wonderful Java based system that implements certain notions that are stated as standards by W3C, and yet the issues of performance and user friendliness have been hard to demonstrate. At least this is many people's experience.
The Second School of Semantic Science feels that the W3C standards has focused too much
· on the URI as a means to provide unique definitions to meaning, and
· too much on the notion that standard logics have to be accommodated in ways the lean themselves to some specific types of "checking".
The Second School has been developed to work within the context of where the ontological modeling community is at today, and to realize a transformation of initial design assumptions by W3C.
Our intellectual positions are expressed in a Preface to a on line book and in a Roadmap that is being discussed in Industry as an alternative to the design philosophy so far stated by the W3C.
I have talked, briefly, with Eric Miller and Tim Berners-Lee and find that they are interested in the second school's viewpoint, but I will not speak for them. I am sure that they are open to a by-pass to what certainly appear to be long term problems with the W3C standards.
The JENA developer’s forum is dedicated to helping people with the JENA versions.
I do not want to occupy this forum with what many might regard as a philosophical discussion.
But we feel that technology is available that has performance of the type needed in bioinformatics and in total information awareness systems like the one I have designed for US Customs.
I thank the moderator for allowing this note to be posted.
Dr Paul S Prueitt
and invite JENA-developers members to communicate to
http://groups.yahoo.com/group/ontologystream/
as well as to being part of the JENA forum.