| ||||||||||
| | ||||||||||
| | The GGG or Puff, is the magic gone? - The Semantic Web on Online International, LondonPublished on 2009-12-15 GGG is Tim Berners-Lee’s acronym for Giant Global Graph, an expression for the semantic web which can be seen as the step that follows WWW. All data on the Internet related via agreed metadata schemas would make it possible using machines to extract new knowledge. This will benefit everything from advanced research to the “best” method to cook Paneer Palak (Indian cheese and spinach). We know that html is a presentation markup, xml a definition markup and RDF, RDF schemas, tell us about the content of the document with the aim of identifying resources. It is done by expressing the relationship of entities. In practice RDF embeds xml in a html presentation. Most of us do not have to understand exactly how this is done. As librarians and being part of the information society we must however evaluate this in respect to our services. How can this add to a better experience? Content does not matter if we do not get it. Here we have a method which implies that not only will we get it, it will also extract the “most relevant” chunks and abstract new levels of information. This new information will be adapted to the context that I am based in when making my enquiry. And even better, I do not express an enquiry, I get serviced just by moving around on the web. I get efficient information on levels that I did not know I could comprehend. On a basic level it is “just” all things connected via relationships (not “simple” links). An example: If you want to buy the book “Sirk on Sirk: interviews with Jon Halliday”* and you fill in a form in one service on the web, theoretically all buying and selling services on the web would thereby know about your wish and could address you with an offer. To make this happen all services need to have the same metadata descriptive framework and there need to be applications doing the connectivity and keeping account of the relationships. Tim Berners-Lee described that all of the Internet wouId be like one database. If we have this then it would also be possible, on the next level, to make calculations on the content and the relationships to extract new information. With these algorithms we would then have the semantic web. One of the main tracks of Online Information in London this December dealt with the semantic web. Most of it was about how it worked and the “fact” that it is the next thing to come. Today on conferences like this we also get practical evidence of that it is, in reality, working. This is done in smaller and larger projects. I may be wrong, and to be frank I am not competent enough to make a real judgement, but as I understand it, it seems that the more limited in scope the project is the easier it is to implement the semantic metadata logic and get results out of it. One presenter, Paul Sonderegger from Endeca, did mention some questioning about the semantic web and said that machines are able to calculate evidence but you need people to judge relevance. Clay Shirky (who Paul mentioned) wrote an interesting article about the semantic web. If I paraphrase him with the example above we will have the following: Boris wants to buy Sirk on Sirk: interviews with Jon Halliday, which gives Boris is interested in books by Jon Halliday Jon Halliday has written a book on Mao People interested in Mao are interested in communism Boris is interested in communism (Boris is a communist) Of course this is far-fetched. But not for a machine building relations on schematic metadata. We see here one of the challenges for the semantic web, the integrity question. Please read the article by Shirky, it is interesting. Maybe the semantic web should be questioned as a solution for the whole web and should instead, as it is now, be used in clearly defined areas. Semantic technology? In the product BookDesire qualitative subjective metadata is added to the catalogue record (also q data) to make it possible to refine the search for a good book. With advanced algorithms on the relationships we will find the book which is closest to our wishes. I want a very long mystery book about the rural eighteenth century with a lot of humour and love. There is none with a lot of love so instead the product will suggest a novel with some love as the first alternative. This service would have been easier to build with a filtering technique but that would not have given me a result. Agreed and defined qualitative metadata cooperates via algorithms. This is value based semantics and an example of that it can actually work. So we, and others, will continue to experiment and implement technologies handling relationships of metadata. A very foreseeable thing to do for a library technology vendor! We would like to discuss areas and projects that you believe will have a valuable impact. *Douglas Sirk was a very fine director whose movies ( I´ve seen a dozen) are no less than a fantastic enjoyment. Comment on “Puff, is the magic gone”: “Puff the magic dragon” was a Peter, Paul and Mary hit which was analyzed beyond proportion. I am just playing around with the notion that the semantic web hype maybe is out of proportion. Work in projects and time will tell.
|
| ||||||||||||||||||||||||||||||||||||||||||
