Why the Semantic Web does not work in 2005

Kerstin Zimmermann
kerstin@physik.org

From an information management perspective we focus on academic and research for the status quo. To see what’s really going on let me start with some historical remarks.

History

First mentioned in 1989 by Berners Lee the Semantic Web became a hype around 2000 within the new economy. EU fundings and US activities spend mill of € and $ in international projects. Data with semantics should be generated automatically in order to create knowledge. The same approach was done in artificial intelligence. But then the dot bubble burst without much content.

So the interests went back to fundamental research of the basics for the Semantic Web. Logic is used to describe things and a very formal way. New languages were developed like OWL for ontologies in the US and later WSML started in Europe for web services. The academic discussion for Web Services went on using always the same example of ordering and buying a train between countries. Companies do not see the market in this application because they are not really integrated and there are no competitive providers in this field because every country still has one national railway.

The add on value of a worldwide semantic network can at the moment only be demonstrated in small parts because of the lack of ontologies. There are no real ontologies or semantic (web) services in use. You can only find island of enhanced databases or prototypes. Although real case applications are not out there although everybody would like to have or at least take a look at these ontologies. Another approach is semantic annotation of named entities on 'normal' web sites like KIM does. But like in all commercial applications the scheme behind is not public. The user only stays with the front end.

What are the problems regarding the full potential of online information? This depends on the conceptual model everybody has in mind when taking about ontologies. In interdisciplinary research areas and applications people have different understanding of them. First we have to specify the model used in computer science. Although here terms are used different in literature like concept and relation in logics but class and attributes for domains. Keeping the history of ontology in philosophy with its fundamental categories in mind everybody can estimate the confusion among different scientists not only talking about pure language coding.

But for a real world application user and expert have to meet and exchange knowledge otherwise ontologies will not being spread around in the future. Developers also have to reflect their background, because every classification schemes the same as in librariy science include implicit knowledge and rankings.

Parallels

The activities to make the Semantic Web real remind very much on the introduction of plain Dublin Core (DC) meta data in the mid 90ies. The format and the technical procedure were quite clear but not wide spread until the first interdisciplinary projects like Dissertations Online in Germany started. Here the workflow for the documents as well as the user as author and information seeking person were included. The task force was first distributed among the expert within the project and for international cooperation, later among the participants when the national library took over the results afterwards. Information management gained add on value for all players.

At the moment an integrated project for semantic libraries has started at DERI called MarcOnt . In one part the wide spread MARC21 bibliographic description format is mapped to the DC elements and the LaTeX format wellknown as scientific publishing tool within a meta ontology.

Ontologies with instances form a knowledge base which is similar to relational data bases with explicit schemes. Creating an ontology with classes and subclasses and the inheriting of attributes can be compared with object- orientated programming and their properties. New classification schemes and multi-lingual thesauri are developed in many new areas, some are although related to libraries see e.g. the Thesaurus for Life Science published in 2004.

Intellectual input is needed because humans think rather different than machines can do. Members handle things within their community in a flexible structure. Depending on the context we have not only to cross the cultural divide between industry and academics but also between different cultures like the Western and the African part of the world. Here a great advantage can be to start from scratch with different concepts of traditions.

Future aspects and Conclusion

Realizing the Semantic Web can not neglect the human factor and has to taken the community aspect into account. Different applications have to be built with real ontologies for specific areas and meta ontologies for the global context. This knowledge has to be made explicit at/to all this also includes a paradigm shift from computer science to library and information science and backwards. Only in this way the methodology can be transferred to different domains.

So my estimation for perspectives of common use are maybe 2010. We'll see what's out there then!

References

[1] Michael Wiesmüller, Reinhard Goebl, Helmut Gassler, Reinhold Hofer, Wolfgang Polt, Wolfgang Resch, Thomas Zergoi, Kerstin Zimmermann (ed.)
Information and Communication Technologies - A Handbook in the Austrian Research and Innovation System
bmvit, Vienna, March 2006

[2] Jiri Panyr
Thesauri, Semantische Netze, Frames, Topics Maps, Taxonomien, Ontologien - begriffliche Verwirrung oder konzeptionelle Vielfalt?
Information und Sprache, Saur Verlag, München 2006

[3] Kerstin Zimmermann, Julika Mimkes, Hans-Ulrich Kamke
An Ontology Framework for e-Learning in the Knowledge Society in Knowledge Organization for a Global Learnung Society
Proc. of 9th ISKO Conf., Vienna, July 2006, pp. 85-92

[4] Sebastian Ryszard Kruk, Marcin Synak, Kerstin Zimmermann
MarcOnt Initiative - Mediation Services for Digital Libraries Ontology Mediation Patterns Library
ECDL2005, Vienna, September 2005

[5] Sebastian Ryszard Kruk, Marcin Synak, Kerstin Zimmermann
MarcOnt - Integration Ontology for Bibliographic Description Formats
DC2005, Madrid, September 2005 Jos de Bruijn, Douglas Foxvog, Kerstin Zimmerman Ontology Mediation Patterns Library D4.3.1 SEKT Project, DERI, February 2005