Home › Forums › SIG: Content/Collection Manager ›
named entity extraction + disambiguation?

named entity extraction + disambiguation?
Posted on: Mon, 05/19/2008 - 07:24
onpause
User offline. Last seen 1 year 42 weeks ago. Offline
Joined: 12/31/1969

can calais tell thedifference, currently, between the concepts behind:

http://en.wikipedia.org/wiki/Cambridge%2C_Massachusetts

http://en.wikipedia.org/wiki/Cambridge 

 

?

 it seems to assign the same hash GUID to all "cambridge"'s

 

best--

 

--chris sizemore 

 

 

Trackback URL for this post:

http://www.opencalais.com/trackback/483

Login or Register to post a comment.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.
Re: More disambiguation woes
Posted on: Thu, 04/02/2009 - 10:26 #1
cwarner
User offline. Last seen 2 weeks 4 days ago. Offline
Joined: 03/31/2009

 So I was playing around with OpenCalais a bit and fed it this url bumped up my threshold to 0.5 and got the following reponse for entities:

City - New York

Person - David Paterson

Person - Sheldon Silver

Person - Sheldon Smith

Currency - USD

Company - Intel

PublishedMedium - New York Magazine

So all of that is correct except for Company. Essentially the link in question refers to Daily Intel a weblog.

HTML content
Posted on: Fri, 04/03/2009 - 08:49 #2
Ofercalais
User offline. Last seen 4 weeks 5 days ago. Offline
Joined: 12/31/1969

Hi try SemanticProxy. This service tries to clean all the non relevant info from the HTML article. (At the moment there is a bug with how the XML/RDF output file is translated into HTML (http://semanticproxy.com/demo.html). We will fix this soon. However the XML/RDF has the right information inside (http://semanticproxy.com/usingSemanticProxy.html).

In future versions of Open Calais we will add this HTML cleansing capabilities to the api.opencalais.com.

 

Ofer 

 Yeah this looks like it
Posted on: Fri, 04/03/2009 - 10:51 #3
cwarner
User offline. Last seen 2 weeks 4 days ago. Offline
Joined: 03/31/2009

 Yeah this looks like it cleans up pretty well for the entities I posted and it doesn't even show Daily Intel which doesn't matter to me in this case. I'll just wait till it hits the API then..

Geo disambiguation
Posted on: Tue, 10/28/2008 - 05:10 #4
MichalF
User offline. Last seen 18 weeks 14 hours ago. Offline
Joined: 05/16/2008

Hi Chris -

Geo disambiguation has been released as a new feature in R3.1.
R3.1 is currently in Tech Preview (read the blog entry for more details) and it will soon be rolled into production.
Try to experience with the new functionality and let us know if it solves the identity problems.

Regards.

Cambridge..
Posted on: Mon, 05/19/2008 - 23:45 #5
Tom
User offline. Last seen 1 week 1 day ago. Offline
Joined: 05/07/2008

You're right. This is a weakness right now. Calais understands City, State, Country, etc - but doesn't understand the combination of these as a unique item. It's something we're thinking about (see other posts on address extraction). and want to make happen in the reasonably near future.

Paris, TX and Paris, France would be a startling mistake for a traveler to make,

Regards,