Path: utzoo!utgpu!water!watmath!clyde!att!osu-cis!tut.cis.ohio-state.edu!bloom-beacon!G.BBN.COM!MVILAIN From: MVILAIN@G.BBN.COM (Marc Vilain) Newsgroups: comp.ai.digest Subject: Case memory for a case-based reasoner Message-ID: <19880718042745.0.NICK@HOWARD-JOHNSONS.LCS.MIT.EDU> Date: 18 Jul 88 04:27:00 GMT Sender: daemon@bloom-beacon.MIT.EDU Organization: The Internet Lines: 37 Approved: ailist@ai.ai.mit.edu Date: Thu, 14 Jul 88 16:48 EDT From: Marc Vilain Subject: Case memory for a case-based reasoner To: ai-folks@G.BBN.COM BBN Science Development Program AI Seminar Series Lecture CASE MEMORY FOR A CASE-BASED REASONER JANET KOLODNER Georgia Institute of Technology, & MIT (AI in Medicine Group), & Thinking Machines Corp. (janetk@zermatt.lcs.mit.edu) BBN Labs 10 Moulton Street 3rd floor large conference room 10:30 am, Tuesday July 19th *** NOTE: NOT THE USUAL ROOM *** Perhaps the most important support process a case-based reasoner needs is a memory for cases. Analysis of observations of physicians using cases during problem solving have led us to derive requirements for a case memory. We then created representations, retrieval algorithms, and selection heuristics that support these requirements. In this talk, I first present observations of physicians using cases during problem solving and then present the requirements on memory that arise from analyzing doctors' behavior. I will also present the representations, retrieval algorithms, and selection heuristics that derive from those requirements. The memory model is implemented in a computer program called PARADYME (Parallel Dynamic Memory) and runs on the Connection Machine. Research was done in conjunction with physicians at New England Medical Center and programmers at Thinking Machines. -------