Slashdot Mirror
C|Net Integrates Ontology Viewer Into News Site
ikewillis writes "The new beta version of news.com now features an integrated ontology viewer developed in collaboration with LivePlasma who appears to have built a large ontology for music and movies. While they don't appear to provide direct access to the ontological data using semantic web formats like OWL and RDF, it's the first time I've ever seen web ontologies used on such a high profile site. How long until we can expect web ontology viewers (and semantic web integration) for sites like Wikipedia?"
I've noticed, for example, that the "Macintosh" (the computer) section of Google News often has non-Macintosh-related stories about sports, crimes, political events, etc. just because a person named "Macintosh" was named in the story. Smarter semantic analysis of news stories would help better categorize articles.
Two wrongs don't make a right, but three lefts do.
For the uninformed, here is some valuable information about ontology.
To get an actual working version of this thing, you have to go to the beta news site and then click on any of the story headlines.
And now I'm sitting here with a room full of sticky webcams!
I guess I just came at this from the wrong angle.
The baby's fine -- please stop sending business cards.
Maybe the poster was looking for something like that.
There are many area where an ontological search (not necessarily graphical like C|Net) is very useful. For example. I started writing a search engine for medical texts which used a medical ontology underneath. It made it so you could search for "heart attack" and get back results about "myocardial infarction" which never mentioned the term "heart attack."
An ontology can make your search much better.
J'aime mieux les méchants que les imbéciles, parce qu'ils se reposent. -- Alexandre Dumas
For instance, as useful as Google is, it's a pain to try to perform queries for things such as "a disease that begins with the letter 'c' and involves a body's inability to produce energy from flour-based foods". With an ontology-based data source, one simply needs to write an interface that allows the user to construct such queries using a formal grammar:
etc. that's just one possible example, but semantic knowledge is infinitely more powerful than grammatical knowledge, and ontology is the genesis requirement of semantic webs.Yes, in a world in which 95% of users can't grasp simple boolean modifiers, such a scheme would surely be a success!
TODO: Something witty here...
The semantic web expects everyone to agree on one ontological framework (one master ontology) and further for each and every web page to markup parts of the page (or the entire page) by indicating parts of the ontology which refer to that piece of text. Then a search engine will come along and use the semantic information encoded in OWL (or some other RDF variant) to know what the page is able and to provide better search results.
The problem is that this process puts far far far too much responsiblity on the web page author. First, they must be aware of this obsecure project. Second, they must understand ontologies and markup their pages honestly. Third, they must maintain this knowledge against shifting ontologies, and the drift of human language both geographically and over time.
Ignoring for a second that people tend to spam search engines in the ever increasing competition for hits. Most people don't have the time, expertese or patience to add this information to the page. It will just be used to fool the search bot just like the meta tags that most search engines currently ignore.
There are good WSD (word sense disabiguation) technologies currently being developed that can figure out from context clues which meaning for a specific word is intended by the author. And these tools are generally built around wordnet which is the ontology that most AI researchers use (and it isn't in RDF, OIL, OWL or any of the other stuff from the W3C). AI researchers know the semantic web won't work because of the reasons outlined above and a few more I can't think of right now. Search engines are pretty good and will only be getting better with time. Quit pimping the semantic web. It only makes you look ignorate in the eyes of the AI community.
"Those that start by burning books, will end by burning men."