Home › Forums › Calais Forums › Geek Speak (Technical usage and issues, questions, and hints ›
Entity detection: How much context is good?
Entity detection: How much context is good?
Posted on: Thu, 09/18/2008 - 23:43

Hi, I submitted the following text:

"Smith Barney: Provides brokerage, investment banking and asset management to corporations, governments and individuals.
Welcome to Smith Barney Benefit Access:Smith Barney offers a suite of online applications designed to help you monitor and manage your employee benefit, compensation and/or equity based plans. (business finance)"

and Smith Barney did not get detected as an entity. Any ideas of why this is? I added "(business finance)" as context, following the steps suggested in another post. 

Any lights on how to better frame this entity are appreciated.

thanks. 


Comments

Tough text
Posted on: Thu, 10/09/2008 - 15:53 #1

This is kind of tough text - the way it is structured there are very few clues that Smith Barney is a company.

For example, if the text is changed to "Smith Barney provides...." Calais wakes up and realizes that it is a company.

We'll forward this along to the NLP team to ponder. Perhaps we're paying a little too much attention to colons?