/* ---- Google Analytics Code Below */

Friday, July 17, 2009

Extracting Knowledge from the Web

In the enterprise we discovered how difficult it was to construct all the rules to provide the capabilities required for even a narrow slice of corporate knowledge. We then experimented with a number of methods that allowed us to induce knowledge from enterprise sources. We looked at early knowledge construction methods like Cyc. That was before the Web was common. Now there are new ways to think about this, still not complete, but steps in the right direction ... Making a system learn is the ideal AI.


'From IEEE Computer: Extracting World Knowledge from the Web by Alexander Yates, Temple University
By automatically extracting information from the Web, we can scale up the resulting knowledge bases to much greater sizes than current collections of manually gathered and user-contributed knowledge .... '

1 comment:

Mark Montgomery said...

Good article- always find it interesting that academia assumes that everyone publishes, and so the public is aware of everything that has been done. In the real world that kind of assumption is dangerous.... thanks- MM

"For instance, with
an approximate inference engine,
an extraction system can greatly
improve its ability to judge the validity
of new extractions.
Despite these advances, there are
numerous, largely unexplored opportunities
concerning the connection
between extracted knowledge bases
and AI applications."