Premium
ON A GENERAL APPROACH TO HEDGED REASONING
Author(s) -
Delgrande James P.
Publication year - 1994
Publication title -
computational intelligence
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.353
H-Index - 52
eISSN - 1467-8640
pISSN - 0824-7935
DOI - 10.1111/j.1467-8640.1994.tb00144.x
Subject(s) - citation , categorical variable , computer science , information retrieval , natural language processing , artificial intelligence , linguistics , psychology , library science , machine learning , philosophy
Henry Kyburg, in “Believing on the Basis of the Evidence” (henceforth BBE), presents an outline of a general approach for arriving at accepted statements on the basis of uncertain information. One begins with a set of sentences BK: representing background knowledge, and a sentence E representing evidence. The task is to determine whether a conclusion C follows. The information in BK: is (typically) anived at via a process of statistical inference and so is not known or certain in any logical sense. Similarly the evidence E may have been accepted only at some level of confidence. The goal (at least as I see it) is to provide a justification, and presumably mechanisms, whereby C may be, without caveat, accepted. Further, the justification and mechanisms developed will also be employed in formulating DK: and E from a “higher” level of background knowledge and evidence. This paper then may be taken as proposing a general approach whereby an agent may come to hold (commonsense) beliefs through interaction with an external worlld. There are, essentially, two fundamental forms of inference relevant to the task. These forms are given early in the paper: