Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!mcsun!ukc!servax0!csc2!scotp From: scotp@csc2.essex.ac.uk (Scott P D) Newsgroups: comp.ai Subject: Re: LOOKING FOR AN ALGORITHM THAT CONVERTS RULE BASES INTO DECISION TREES Message-ID: <5046@servax0.essex.ac.uk> Date: 26 Apr 91 07:48:26 GMT References: Sender: news@servax0.essex.ac.uk Reply-To: scotp@essex.ac.uk (Scott P D) Distribution: comp Organization: University of Essex, Colchester, UK Lines: 21 In article jasko@charles.bu.edu (John V. Jaskolski) writes: > I am looking for any information (textbooks, articles, etc.) >regarding algorithms that convert rules into decision trees. I am >aware of papers that describe mechanisms for converting decision trees >into rules (e.g., Quinlan's "Generating Production Rules From Decision >Trees"); however, I have not found anything on converting rules into >decision trees. If anyone can make any recommendations I would >greatly appreciate it. Well, one recommendation might be to stop looking. While there is a set of rules corresponding to any decision tree, the converse is not true. The reason is that trees are (by definition) cycle free, whereas a set of rules can define an iterative or recursive process. Hence, only a rather restricted class of rule sets can be transformed into decision trees. Paul Scott, Dept Computer Science, University of Essex, Colchester, UK.