A few days ago Monika Rani (Research scholar at IIIT-ALLAHABA) asked this question on a LinkedIn group.
Did Linked Data Kill Ontologies? Cited was a panel
discussion entitled "Does the Semantic Web Need Ontologies" and David Karger and Abraham Bernstein of University of Zurich captured in a
Web 3.0 video a transcript ia here: http://www.allreadable.com/62faAA3v .
"Sloppy, scruffy Semantic Web does not need ontologies and that Ontologies are a luxury and should not hinder open data publishing and usage."
There has been some small reactive discussion such as Mitch
DeFelice:" Nope it just validated the separation of concern. Linked Data and Ontologies having two different roles. Linked Data is simple a means of publishing a graph of data, whereas Ontology is the model that sets context from the graphed data."
Barry Zane thought the answer was" Yes and No. The larger job of the ontologies
is to relate inconsistently named things in the data. Since the governments of the world can't even agree on exactly what a "ton" is, it is hard to imagine people supplying data to be consistent with other people supplying data. The best we can hope for is
that each data dataset has an internally consistent (possibly implied!) ontology that a connecting ontology can be layered on. "
I thought that the Ontolog community might have more thoughts on theis issue since both ontology and LPD is spoken here.