[om-list] hello again ...

[MrM0j0r15n Rajeev the kanehbosm]

Hello friends,

In my humble opinion, i think it is impossible to derive immediate (6months 
to a year) results by trying to craete a system that will somehow perform 
constructive generalizations on knowledge contained in documents, using a NL 
frontend. (frankly, i dont think computers can do that at all).

So, i propose to build a system to perform this in a human-assisted way. It 
seems that it is possible to utilize human beings in the form of annotaters, 
to markup documents so that an inference engine may be used on it (please 
see http://www.cs.umd.edu/projects/plus/SHOE/ ). I believe that once enough 
of knowledge has been represented in this way, it will become possible to 
build another system which should derive the same knowledge base from the 
same set of documents. The possibility arises because, now it will be 
possible, to elegantly test such a system with the previous system that we 
have built.

The essential question of whether a knowledge system like OM will work, 
rests on the correctness of its inferences (from the established set of 
documents). A human created knowledge system is, i think, the best bet for 
this test. Now, it will also be possible, i think, to create bridges between 
the systems, so that only a few (comparatively) human editors will be needed 
to correct the system in case it makes errors in its NL inferences.

plss visit the URL mentioned above.

XSB Database - http://xsb.sourceforge.net/
ALICE - http://alice.sunlitsurf.com/

insanekane

PS: i dont whether i should shutup, or keep talking. pls enlighten me


_________________________________________________________________
Send and receive Hotmail on your mobile device: http://mobile.msn.com