Re:Scientific publications should be anonymous
by phreeza on Sunday December 20, 2015 @05:47AM
(#10766705)
> How is a search engine supposed to make those connections?
I can't tell you how exactly, but if current word embeddings can do stuff like "Paris - France + England = London", I don't see why future systems shouldn't be able to resolve the relationships between institutions, people and publications you mention, in a similar fashion. I am not saying that it will be easy, your example requires much more "one-shot learning", but I could in principle see how an automated system might do it.
I can't tell you how exactly, but if current word embeddings can do stuff like "Paris - France + England = London", I don't see why future systems shouldn't be able to resolve the relationships between institutions, people and publications you mention, in a similar fashion. I am not saying that it will be easy, your example requires much more "one-shot learning", but I could in principle see how an automated system might do it.