Semantic speech

Until I joined this group last year, I hadn't touched SR since my brief stint at SRI International about 15 years ago. I was disappointed to find that in the intervening time very little had been done on using semantics, i.e. word meaning, as a way to enhance the quality of the recognition. So I was intrigued today when I heard about Logical Semantics, Inc., a company spun out of Indiana University a few years ago to do just that.

I wonder how the core technology works. Is it rules-based? or statistics? One of these days I'll study their web site more closely to understand.