maybe there is a general miss definition in topic maps environment. I read that topics per definition are independent from any context (Closs, Sissi 2007)... where most semantic web approaches in my option target on creating more context between given terms.

Topic maps may be good thing for cross media publishing.. having one information only in a single instance and resusing it ... but for semantic issues the context counts more. and maybe a term cannot stay alone without connections around it in many situations (e.g. the linguistic problem of homonyms)

Closs, Sissi (2007): Single-Source-Publishing. Topicorientierte Strukturierung und DITA. [Frankfurt (Main)]: Entwickler.press. (german)

Uh.. I've given the old book as a source.. there is a newer one:

Closs, Sissi (2011): Single Source Publishing. Modularer Content für EPUB & Co. Frankfurt, M: Entwickler.press. ISBN: 9783868020786

Thanks Andrè Kleinschmidt for the comment and the reference of the book.

What about reasoning on Topic Maps? While semantic web do reasoning and verification TM do only verification, but is it possible in your opinion to do reasoning too? How can it be done, do I have to trasform XTM into RDF ?

Thanks in advance.

Hi @Frosina and all, actually I am also interested in finding out if anyone knows about reasoning mechanisms in TMs. I am not sure there are appropriate reasoning engines PLUS the transformation to an RDF/OWL model is not loss-less, so it is not the ideal solution.

Efstratios, Frosina, this is a widely discussed topic. In fact, Frosina points into the right direction: where RDF/OWL do have reasoning and verification mechanisms, Topic Maps do only have verification mechanisms built in (Topic Maps Constraints Language, TMCL). Some time ago we built an implementation of TMCL (http://www.topicmapslab.de/projects/tmcl_validator). But within the Topic Maps Query Language (TMQL) there are some basic reasoning functions built. However, TMQL does not provide the full functionality which is usually expected by referring to "reasoning".

In general, it is always worth to think about the potential use cases for your knowledge graph. For the most use cases, the funcationality of TMQL is absolutely sufficient, but for others the reasoning features coming with OWL might be necessary.

Regarding to the argument " the transformation to an RDF/OWL model is not loss-less" it might be important to understand what kind of information is lost by making the mappings. If you transform Topic Maps Data into OWL you might loose some Topic Maps specific semantics (which come from the Topic Maps Data Model, TMDM). But I would argue, that this kind of information is not necessary if you want to apply the OWL mechanisms on the data which originally cam from topic maps. In the most cases you might be able to apply such an approach.

But obviously, if you make a round-trip transformation TM --> RDF/OWL --> TM you will usually loose some information and the new topic map will probably not have the same caonoical representation (CXTM) like the original topic map.

I will argue that the OP's premise about "resurrection" of topic maps can be viewed as false.

In some cases, it is as simple as funding drying as in the SemanticWeb project changing the funding landscape.

It's true that some of us built topic maps in RDF and applied some logic engines to them. But, some of us have seen a fundamental impedance mismatch between the topic mapping and SemWeb architectures and goals.

Never the less, while topic maps are far less visible, they are alive and well, with no need for resurrection.