On the automated construction of KR

following on the earlier exchanges

 automating knowledge extraction to derive a representation (rather than
building it from the top down) requires that
- the universe of discourse from which the graph (awww) is derived is
correct-
and/
- the parsing of the knowledge is carried out correctly
and we need some KR to do evaluate it other way
I dont think we can get away from using KR

PDM