Accosted By A Barker

When you submit papers to academic conferences, then you have to be prepared that some of them will be rejected. And this is all good and well.

Now, in the years 2004 to 2006 I have submitted numerous papers to Semantic Web related conferences, such as the WWW conferences or ISWC. Some of them with significant effort.

They have all been rejected, often with completely empty reviews, but more often with yes, you have done a lot of work but why are you using TMs and not DTD or RDF or OWL or SPARQL?.

And, no, it does not matter whether you confirm with the program chair, that papers using TMs are suitable for that conference. And, yes, it is infuriating to get shot down a formal optimization calculus for TMQL evaluation which took you months of your life with a review like the above.

So the only place to take TM-related research is TMRA, a conference which I normally enjoy a lot.

But RDF zealotry seems to have reached TMRA too, as the following review of a paper about a "Topic Map based File System" indicates:

This paper handles only RDF, but W3C already recommended OWL (Web Ontology Language). I mean authors first define differences bewteen OWL and TM and clearly show why their proposal is based on TM.

To be honest, I have some reservations why I should motivate the use of Topic Maps for a conference dedicated to Topic Maps. But probably I'm just stubborn. Again.

In the detailed comments for the authors the RDF Mullah then continues to wag with his index finger:

This paper mentioned the RDF model offers only the construct of a triple, such mapping is rather straightforward. ... W3C also provides an mechanism for interoperability between RDF and TM, and so there are many translation algorithm between them. It means RDF covers all vocabularies or meanings provided by TM. Finally, W3C developed and recommended OWL (Web Ontology Language) which is more powerful than RDF or RDF Schema. However, this paper didn't handle it.

Not even Bond University students did say or write such gibberish. At least not for long. And, no, the comment has nothing to do with the contents of the paper either.

Consequently, a more deep survey is required to show the originality of this paper.

Sure. Thanks, but in the meantime I'll listen to London Town:

Walking Down The Sidewalk
One Purple Afternoon,
I Was Accosted By A Barker
Playing A Simple Tune Upon His Flute.


Posted In


Do you think the RDF community tries to undercut Topic Maps because Topic Maps is the better technology? ;)

Lars (not verified) | Sun, 08/31/2008 - 14:13

Re: undercut Topic Maps

At this stage of the development, the question which technology actually is better is pretty irrelevant. And even if they were better, I cannot remember any incident in history where this was significant in a competition between technologies.

TMs are not taught in universities, therefore students do not know about them, therefore there will be hardly new open source software or external dynamics. Now it is completely in the hand of the commercial wing of the TM community, what will happen next.

rho | Sun, 08/31/2008 - 20:32


Well, a serious answer to a conclusion which was not meant seriously.

Lars (not verified) | Mon, 09/01/2008 - 13:24


was this submission accepted? i remember i saw one event about topic map filesystem at the tmra shedule.

schtief (not verified) | Sun, 08/31/2008 - 17:01

Re: accepted?

Yes, the paper was ultimately accepted by the program committee.

rho | Sun, 08/31/2008 - 20:28

TM File System

A TM-like file system is something I was pondering recently after reading about Semantic File Systems and Logic File Systems:

Do you have an pointers to research or discussions in this area, or was your paper the first? (I'll try to track down your paper too)

Would you construct paths using the TMQL path syntax (maybe via "mkdir") and navigate using shell commands like "cd" and "ls"?

Michael (not verified) | Tue, 09/02/2008 - 07:28

Re: TM File System

The TMFS we have implemented is not a semantic one. It is actually the inverse idea of using a conventional file system to navigate through a topic map:

cat /internet/browsers/mozilla

gives you the whole topic (encoded in AsTMa at the moment) and

echo "http://...." >> /internet/browsers/mozilla/homepage

adds one more URI as a homepage occurrence.

One cool addition (which we have not implemented) is to stack such a file system on top of a real one, so that you can keep your real documents and your topic man concent "in the same directory".


The idea of a semantic file system is definitely worth following up, although you ultimately end up with something which is not "file-systemy" at all.

rho | Tue, 09/02/2008 - 09:44