LEWIS@cs.umass.EDU (11/04/88)
Can anyone point me to some references on matching of subparts of frame-based knowledge representation structures? Essentially what I'm interested in is equivalent to finding some/all/the biggest of the isomorphic subgraphs of two directed graphs, except that edges and vertices are labeled, and there are restrictions on what labels are allowed to match. For additional fun, there might be weights on the edges and vertices as well, and you might not just be interested in large-sized isomorphic subgraphs, but in maximal scoring ones. Still more interesting would be if anything has been done on the case where you can inferences to the structures before matching, so that you actually have to search a space of alternative representations, as well as comparing them. Suggestions? If text content matching had been a bigger application of NLP in the past, there'd be a bunch of stuff on this, but as it is, I suspect that vision or case based reasoning people may have done more on this. Best, David D. Lewis ph. 413-545-0728 Computer and Information Science (COINS) Dept. BITNET: lewis@umass University of Massachusetts, Amherst ARPA/MIL/CS/INTERnet: Amherst, MA 01003 lewis@cs.umass.edu USA UUCP: ...!uunet!cs.umass.edu!lewis@uunet.uu.net