clarke@utcsri.UUCP (04/10/87)
(GB = Galbraith Building, 35 St. George Street) SUMMARY A.I. SEMINAR, Tuesday, April 14, 3 pm, GB120 -- Alan Frisch: ``Knowledge Retrieval as Specialized Inference" SYSTEMS SEMINAR, Thursday, April 16, 11 am, GB220 -- Norman Hutchinson: ``The Implementation of Emerald" ------------- A.I. SEMINAR, Tuesday, April 14, 3 pm, GB120 Professor Alan Frisch University of Illinois ``Knowledge Retrieval as Specialized Inference" This thesis sets out to study the nature of retrieval from knowledge bases. Formal specifications that capture certain informal intuitions about retrieval are developed, studied, and implemented by retrieval algorithms. Seemingly at odds with the intuition that a retriever is a pattern matcher, this thesis views the entire process of retrieval as a form of inference and hence the KB as a representation, not merely a data struc- ture. A retriever makes a limited attempt to prove that a queried sentence is a logical consequence of the KB. When constrained by the no-chaining restriction, inference becomes indistinguishable from pattern-matching. Imagining the KB divided into quanta, a retriever that respects this res- triction cannot combine two quanta in order to derive a third. The techniques of model theory are adapted to build non-procedural specifications of retrievability relations, which determine what sentences are retrievable from what KB's. Model-theoretic specifications are presented for four retrievers, each extending the capabilities of the pre- vious one. Each is accompanied by a rigorous investigation into its pro- perties, and a presentation of an efficient, terminating algorithm that provably meets the specification. This work should be of particular interest at the University of Toronto since it proposes a three-valued logic related to Levesque's logic of explicit belief and extends a technique introduced by Reiter for reason- ing with taxonomic information solely during unification. SYSTEMS SEMINAR, Thursday, April 16, 11 am, GB220 Professor Norman Hutchinson University of Arizona ``The Implementation of Emerald" Emerald is a new programming language that extends the concepts of object-based languages to the distributed environment. In particular, Emerald supports a single, very general model of computation: the object. The Emerald object model incorporates private entities such as integers and Booleans, as well as shared, distributed entities such as compilers, directories, and entire file systems. All Emerald objects are mobile, but invocation is location independent. A major goal of the Emerald design is efficiency. While the object model is very general, this full generality is not required by every object that may be defined. The Emerald compiler accepts the responsibility of providing an implementation for each object appropriate for the level of generality required by it. Emerald is currently implemented on a local network of DEC MicroVax II computers. Performance data from this implementation will be presented to substantiate the claim that Emerald can be efficiently implemented. -- Jim Clarke -- Dept. of Computer Science, Univ. of Toronto, Canada M5S 1A4 (416) 978-4058 {allegra,cornell,decvax,linus,utzoo}!utcsri!clarke