HOME / ¹®¼°øÀ¯ / ¸®Æ÷Æ®/³í¹® / ±â¼ú°øÇÐ
0
0°ÇÀÇ Èı⺸±â(¿µ¹®) Ãß·Ð ¹× ÇнÀ¿¡ ´ëÇØ ±â¼úÇÑ ¸®Æ÷Æ® Âü°íÀÚ·áÀÔ´Ï´Ù.
reasoning with rules production systems (»ý¼º½Ã½ºÅÛ)
rule-based systems (most of the expert systems) are sometimes called ¡°production systems using rules¡± or simply ¡°production systems¡±.
production system consists of fact base, rule base and inference engine (reasoning module).
fact base (short-term memory or working memory): a set of facts rule base (long term memory): a set of rules:
form if <condition> then <results> ¡¦.or actions in <results>.
inference engine: applies an applicable rule to facts (rule triggering,
or rule firing) to find new fact or to perform an action.
-needs conflict resolution (Ãæµ¹ÇؼÒ)-select one rule to apply (depth first, best first, etc.) ex.dfs, bfs, b&b, bestfs,a*
facts rules ie inferred facts actions 53 forward chaining 54
ex.did wilma love barney?
state consists of:
man(fred), man(barney),woman(wilma), spouse(fred,wilma), friend(fred, barney)
rules are: if man(x) or woman(x) then person(x)
if person(x) and spouse(y, x) and friend(y, z) then ~love(x, z),
¡¦ steps
woman(wilma) applied to first rule ¢¡ person(wilma) person(wilma) and spouse(fred,wilma) and friend(fred, barney) applied to second rule ¢¡ ~love(wilma, barney)
backward chaining 55 example ~love(wilma, p) ¡è z/p person(wilma), spouse(y,wilma), friend(y, z) ¡è fred/y person(wilma), friend(fred, z) ¡è barney/z person(wilma) ¡è woman(wilma) ¡è none (the goal is proven)
-interpretation using substitution list: ~love(wilma, barney)
non-monotonic reasoning (ºñ´ÜÁ¶Ãß·Ð)
sometimes, old fact needs to be taken back.
wet(grass) and if rains(x) then wet(x), we concluded rains(grass).
but later, find sprinkler-on(grass), then we need to retract the fact
rains(grass).
in this case, production system needs to perform non-monotonic reasoning.
to perform non-monotonic reasoning, production system sometimes maintains a tms(truth maintenance systems) to maintain truth values for all clauses in working memory, instead of retract the facts.
ex) rains(grass)[0] where [1] true, [0] uncertain, [-1] false
56 handling uncertainties 57
in reality, almost everything is uncertain.
especially in natural and social systems. ex (ÀÌÇÏ »ý·«)
¹ÞÀº º°Á¡
0/5
0°³ÀÇ º°Á¡
¹®¼°øÀ¯ ÀڷḦ µî·ÏÇØ ÁÖ¼¼¿ä.
¹®¼°øÀ¯ Æ÷ÀÎÆ®¿Í Çö±ÝÀ» µå¸³´Ï´Ù.
Æ÷ÀÎÆ® : ÀÚ·á 1°Ç´ç ÃÖ´ë 5,000P Áö±Þ
Çö±Ý : ÀÚ·á 1°Ç´ç ÃÖ´ë 2,000¿ø Áö±Þ