« Info Viz Links From Class | Main | Geographic Data Maps Mashup »
Reuters’ Open Calais
By Nick | February 13, 2008
Reuters release an new API called Calais based on a semantic entity extraction engine they acquired from Clear Forest last year. It can extract Entities, Organizations, Companies, Events, and relationships between these things. Read Write Web posted a good summary of the API’s capabilities here.
Thinking about how to integrate the capabilities of this API into news analysis or consumption interfaces would an excellent starting point for a final project.
Topics: Interesting |
3 Responses to “Reuters’ Open Calais”
Comments
You must be logged in to post a comment.
March 5th, 2008 at 11:47 pm
Although the idea of semantic web seems to be glorious with its countless possibilities to retrieve and understand information and knowledge, there are some caveats with the data available online, especially on private websites, blogosphere etc, which is unfiltered and extremely noisy.
One of the people who has written against semantic web (actually metadata) in 2001 was Cory Doctorow in his rather humorous essay, Metacrap: Putting the torch to seven straw-men of the meta-utopia which is available on http://www.well.com/~doctorow/metacrap.htm
March 11th, 2008 at 9:11 pm
There is a website called FreeBase, (http://www.freebase.com ) which is an open shared database of world’s knowledge.
There are schemas created for different types of entities (if you don’t find a schema for your required entity, you can always make one). Like wikipedia, it can be edited by users of the site. The content can be imported and the API can be used to perform entity extraction (using their Metaweb Query Language http://www.freebase.com/view/freebase/api).
It can be a useful source, if the entities you are trying to look for are not present in opencalais database.
March 23rd, 2008 at 6:02 am
Tom Tague here - leader of the Calais initiative at Reuters.
While Calais is a general purpose metadata generation tool - where we get personally excited is the idea of its application to news.
We’d encourage anyone interested in the next generation of news to experiment with Calais - but by no means should you stop there. The generation of metadata around a specific article is interesting, but what is significantly more interesting is using that metadata as your passkey into a much larger world of relevant content.
Process an article with Calais, look up the results in Freebase or other machine-accessible sources, look for interesting patterns or linkages and create innovative ways to present all of this to your readers - that’s the challenge.
Regards,