ScienceOnline09: The Semantic Web in Science
 John Wilbanks |
John Wilbanks, Vice President of Science Commons gave a fairly technical overview of the Semantic Web, its promise, and what needs to be accomplished for it to fulfill its promise. I’ve been extremely excited about the Semantic Web for a very long time, too long a time. It concerns me that I’m not seeing the benefits of this idea yet: an Internet packed, not just with links between different data, but information about the links themselves.
Part of Google’s page-ranking algorithm concerns how many web sites are linking to a particular site where a keyword appears, known as the “Bag of Words” paradigm. The problem with this approach is that we don’t know why these other sites are linking to this first page. The Resource Description Framework offers a means to correct this oversight, providing a description of the type of link being used:
 RDF-Triples Credit: W3C |
Every arrow in the above example represents an RDF Triple: subject, predicate, object. In the World Wide Web, with every link we all ready have a vague concept of subject and object, but With this addition of predicate, we have the ability to triangulate links into concepts.
Imagine a search engine that knows when you type “primate,” it should also look for “monkey,” “ape,” and “chimpanzee,” or a search engine that knows when you enter “wood,” that you are looking for forestry information and not carpentry. That is the revolution at which the semantic web hints, but never seems to manifest.
Why is that? One point brought up in the session is the lack of Semantic Web software. Currently Nepomuk is, to my mind, a leading contender with its beta Semantic Desktop, but anyone who can code an effective WYSIWYG semantic web tagging tool would strike it rich. In a few years, when I intend to be pursuing a doctorate in Computer Science, I want the Semantic Web to be my focus. There is simply so much to do in this field.
And why isn’t everyone working on it? The fact is, there is too much data online for people to go back through everything all ready out there and add all of this metadata and not enough incentive. It’s like the current attempts to switch to hydrogen cars. There are no hydrogen fuel stations, because there aren’t cars demanding them, and there won’t be cars without the fuel stations. We won’t have a semantic web until we have a web coded with semantic data, and people won’t have the incentive to code in semantic data until there are popular tools making use of that data.
Interesting note: yesterday I mentioned Star Trek’s influence in science, well, for the semantic web, there is a Star Trek Ontology, which is used for hypotheticals where real world examples would be too complicated.
The NeuroCommons is one example of the semantic web in action in science. Wilbanks encourages people to download it and create their own versions as it is open-source.
Note about why I look forward to the Semantic Web: I experienced an example of a “Google FAIL” yesterday, when doing a search for hardware requirements for a file sharing server. No matter how I manipulated the query, most of the results were for peer to peer software. Finally, the following query managed to cut out most of the P2P results:
file sharing server hardware requirements -p2p -”peer to peer”
However, I still didn’t get back much by way of useful data.
Comment by ideonexus — January 25, 2009 @ 2:09 pm
[...] De qualquer maneira… essa “simples” idéia — de assimilar os “metadados” de forma fundamental e intrínseca nas entranhas da Web — tem um enorme potencial quando o assunto é Publicação Científica. Um exemplo claro disso é o Scientific Publishing Task Force: Mindswap: Science and the Semantic Web, Science and the Semantic Web (PDF), Semantic web in science: how to build it, how to use it, ScienceOnline09: The Semantic Web in Science. [...]
Pingback by Matemática na era da Web2.0… « Ars Physica — February 25, 2009 @ 10:10 pm