TopicMaps in Oslo
March 17th, 2008 by David Weinberger
April 2-4, I’m going to TopicMaps, a conference that may be particularly interesting (to people who are particularly interested in it, of course):
The basic idea is simple: the organizing principle of information should not be where it lives or how it was created, but what it is about. Organize information by subject and it will be easier to integrate, reuse and share – and (not least) easier for users to find. The increased awareness of the importance of metadata and ontologies, the popularity of tagging, and a growing interest in semantic interoperability are part and parcel of the new trend towards subject-centric computing.
The organizers have let it be known that there’s still room… [Tags: conferences topicmaps oslo everything_is_miscellaneous]
As far as I am concerned, if every topic in a document can be represented at its meaningful atomic level, maybe in a form of noun phrases, named entities, subjects, etc., then any application can be developed to wrap around the topic, and consequently make the topic reusable and sharable in a pervasive and distributed computing environment.
In other word, every topic can be designed to function as a little distributed computer itself. That is why I was fascinated by the concept of semantic Web and its associated Web standards and technologies back in the late 90’s.
As far as semantic interoperability in subject-centric computing is concerned, I really think that the defined top two-tier subject categories shall represent the logical axiom of a knowledge map that is domain specific, community-centric, process centric, user centric, and independent of any technology.
Ideally, someone or some organization has to provide a Web platform for the agreed upon top two-tier subject categories to be securely defined, developed, managed, reused, shared, improved, controlled, and governed at any point in time when the aboutness of a document is derived by the creator or application.
In addition to supporting semantic Web standards and technologies for the “ontology” or “defined top two-tier subject categories”, a built-in data quality control process for the platform has to be in place with a little touch of statistical and NLP (Natural Language Processing) techniques from the design environment, to build, running, and operational environment.
Thanks!
Amanda Xu
Subject is not context-less — who created it, where and when they did so, are key to understanding. Even an ontology has a domain — there is no universal knowledge representation. And, you cannot extract everything from the content of the object itself (even if it is text, is structured, and has metadata).
Here’s a challenge — try to capture context and its relationship to meaning, instead of trying to separate them.
The TopicMap XML-standard is mainly used by organizations to create closed vocabularies for classification of web pages. The standard itself facilitates creation of advanced ontologies. Since there are few (open source) software implementations of the standard and current TopicMap software is not particularly user friendly, it ill suited for for open user created ontologies or tagging systems. This is because a standard aimed at creating hierarchical ontologies with bi-directional links adds a lot of unnecessary complexity to (flat) tagging systems.
You will probably stir some interesting discussions on the conference, since your latest book is a 233 page critique of what TopicMaps are about to solve, i.e. how to create “nothing is miscellaneous” ontologies :-)
@Lisa: The Topic Maps standard use the term ‘scope’ for context. It is unfortunately not widely used yet, but means that it is possible to take the step from semantics to pragmatics (ref formal language theory/philosophy and syntax-semantics-pragmatics).
@Harald: The Topic Maps standard (ISO 13250) is not an XML-standard. The standardized interchange format for Topic Maps, XTM, is of course an XML-standard, but the underlying model is not. It is also not particularly aimed at modeling hierarchies, but general ontologies, even if I think most implemented systems have used some form of a taxonomy in their ontology. – It’s basically the way humans think.
The article “Metadata? Thesauri? Taxonomies? Topic Maps!” gives a quite good overview:
http://www.ontopia.net/topicmaps/materials/tm-vs-thesauri.html
As for David Weinbergers keynote at the conference, – it was great!
Planet Topic Maps is a good place to look to see what’s happening in the Topic Maps community blogosphere:
http://topicmaps.bouvet.no/planet/
[…] public links >> topicmaps Comment on TopicMaps in Oslo by Are D. Gulbrandsen Saved by GirlNextDoor135 on Mon 29-9-2008 Topic Maps at the Expert Center for Information […]