daniel@bnr-di.UUCP (04/11/85)
Does anyone have any experience with the work of R.S. Michalski (U. of Illinois) on concept learning? The problem I'm faced with is, from another point of view, that of constructing natural responses to database queries. On executing the query, we will get a (possibly empty) collection of values from the database that constitute the answer to the query. Rather than report all 144 elements in the answer, for instance, we would rather summarize the contents of the 144 elements by looking for regularities among them. (In actual fact, this won't be a natural language output; we need something formal for further processing. But I don't think that's important at this point.) I'm aware of some material on natural language responses to database queries, but hadn't come across the "learning" approach of Michalski until recently. Any comments, or pointers to other approaches to the problem? Thanks!! Daniel Zlatin Bell-Northern Research ...utzoo!bnr-vpa!bnr-di!daniel