Internet commentary

Citation

Andrew, A.M. (2003), "Internet commentary", Kybernetes, Vol. 32 No. 7/8. https://doi.org/10.1108/k.2003.06732gag.001

Internet commentary

Keywords: Semantic Web, Search engines, ERCIM, Professor Edsger Dijkstra

Abstract The Semantic Web is introduced as an emerging development from the World Wide Web, promising enormous improvements in performance and ease of use. Sources of information include a free journal. The death of the computing pioneer Professor E.W. Dijkstra is noted.

Semantic Web

The vast amount of information accessible on the World Wide Web is certainly not arranged for the convenience of access. The penetration of the Web, virtually into all areas of knowledge is only possible, because it has grown without overall constraint or direction, but a consequence is that there is no prearranged structure or indexing system. The difficulty of navigating has been alleviated by the introduction of search engines, but it is easy to see that they are not an ideal solution.

A special issue of the journal ERCIM News is devoted to the next development, referred to as the Semantic Web. Euzenat (2002) uses a simple illustration to show the deficiency of the existing search facilities. He points out that looking for a book about a prolific author, say Agatha Christie, is not easy because a search engine given the name will list the many works that are by Agatha Christie as well as those that are about her, presumably including her autobiography, which is in both categories. If the word “about” is included among the keywords inserted, it will have no effect on the search unless one of the titles happens to include it.

A language devised to improve the search capability is termed resource description framework (RDF) and one of its main features is attention to relational information as conveyed by the terms like “is about” and “is author of”. This is part of the “Semantic Web” initiative, first proposed by the pioneer of the World Wide Web, Tim Berners-Lee.

A valuable Web site, with numerous links, is: http://www.semanticweb. org. This includes links to sites where research from different groups is reported, and a “news” section with notices of relevant scientific meetings. There is an invitation to become involved by founding a Semantic Web community and leading an effort to develop a vocabulary for a specific domain, and anyone with this in mind is invited to communicate with Stefan Decker at: stefan@db.stanford.edu. “Challenges” or suggestions for software that might usefully be developed can also be submitted, and sponsors are sought. Suggestions and offers can be sent to Terry Payne at: terryp@cs.cmu.edu.

There is an opportunity to subscribe to an interest group, run by Yahoo, or to join it, by sending an empty e-mail to: semanticweb-subscribe@egroups.com. and then responding to the invitation that is sent. Anyone who subscribes to the group receives periodic e-mail updates on progress, and members additionally have access to certain group resources and can contribute.

The aim of the developments is to let the semantics of the subject matter on the web be available to the machines, so that much more can be done to assist the user. Work on knowledge-based systems, and hence AI, will play a part, but it is not intended that the Web should become autonomous and independent of humans. According to Berners-Lee and Miller (2002) and Berners-Lee et al. (2001), the new version of the Web will evolve from the old one, and will have the same decentralised character. The main tools already exist, one of them being the language RDF, and the other XML or Extensible Markup Language. The decentralised character means that the system will be incomplete, in a mathematical sense meaning that it may fail to give an answer. Berners-Lee et al. (2001) suggest a rule, reminiscent of Gödel's theorem, to the effect that any system complex enough to be useful also encompasses unanswerable questions.

The language RDF allows the setting up of triples that are like the subject, verb and object of an elementary sentence. The “verb” part indicates a relationship like “is author of”, and types of such verbs are stored as Universal Resource Identifiers (URIs) similar to the familiar URLs or Uniform Resource Locators. As with URLs, new URIs can be defined anywhere by anyone and so the semantics of the Web will become continually more comprehensive.

The Extensible Markup Language allows tags to be attached to hypertext that are not represented in the screen display of ordinary browsers. They are semantic links for machine use.

Scenarios

Both Euzenat (2002) and Berners-Lee et al. (2001) illustrate the expected power of the developments by describing the hypothetical situations. Euzenat describes a software “agent” acting as an efficient travel agent and planning a complex trip involving several modes of transport as well as hotel stays and conference and entertainment registrations, at minimum cost. The choices would be subject to constraints of acceptable hotels with vacancies, eating requirements, frequent-flyer affiliated car rental, and so on.

Berners-Lee et al. refer to a brother and sister, arranging to take their mother to clinics for a series of physiotherapy sessions, where both have full appointment diaries and there is a choice of clinics. The personal “agent” of one of them is able to search for clinics within a stated distance and to verify that they are suitably accredited and covered by the mother's insurance plan. It then tries to devise a schedule of sessions, where each slots into a vacancy at one of the clinics and also into free time in the appointments diary of one or other of the siblings.

Agents

Operation depends on the setting up of software “agents” able to communicate. As well as the personal agents postulated, agents would be set up by hotels and airlines and car-hire companies, and also by clinics and insurers and accreditation authorities. Not all information would be freely accessible, and for example, the personal agent (that of the sister, say) forming the plan for visits to clinics would only be able to access the appointments diary of the brother once he had indicated to his agent that the sister's agent was to be trusted. A “verb” indicating “is trusted by” can be represented as an URI, though in practice, something more complex defining the contexts in which access is permitted would probably be wanted in even the most friendly of family relationships.

The possibilities of the Semantic Web are most readily illustrated in terms of these tasks of a scheduling nature, but more significant applications can be visualised for academic research and discussion and the amassing of human knowledge.

Ontologies

Certain information about the structure of knowledge is needed. For example, it needs to be known that a postal address normally contains a house name or number, street name, zip code, etc. The terms “zip code” and “postal code” should be recognised as equivalent, and “autobiography” must be seen to imply both “is about” and (in reverse direction) “is author of”.

The word “ontology” is borrowed from philosophy and used to denote a facility for the formal representation of relations among terms, usually by a set of inference rules. The effectiveness of the Semantic Web depends critically on the construction of suitable ontologies.

ERCIM News

The journal ERCIM News is published by the European Research Consortium for Informatics and Mathematics, and is very generously distributed free of charge. Details can be found on the Web site: http://www.ercim.org/publication/Ercim_News/. Besides offering the possibility of registering to receive the journal, this accepts requests for back issues that can be supplied in either printed or electronic form. A search facility is also provided. Back issues are held from October 1994.

Professor Edsger Wybe Dijkstra

In the same issue of ERCIM News, the death of E.W. Dijkstra, on 6 August 2002, after a long struggle with cancer, was announced. He was the most famous member of the Mathematisch Centrum in Amsterdam (now Centrum voor Wiskunde en Informatica) and among his many achievements was the introduction of the stacking principle, which was the basis of the world's first ALGOL60 compiler. From 1984 to his retirement in 1999, he worked at the University of Texas in Austin.

Details can be found at: http://www.cs.utexas.edu/users/EWD/, including proceedings of a symposium held in honour of his birthday in May 2000 (occupying 10 MB) and a 25 min video. He was the author of more than 1,300 papers and all of them have been made available online, many of them not previously published. There are also warm tributes to his warm and generous character.

Alex M. Andrew