TY - JOUR

T1 - The role of relevance in explanation I

T2 - Irrelevance as statistical independence

AU - Shimony, Solomon Eyal

N1 - Funding Information:
I wish to thank Eugene Charniak for supportingth is researchb, oth financiallya nd throughs olid advice,a nd Eugene SantosJ r. for fruitful discussionos n the reductiont o linear systemso f inequalitiesT. his work has been supportedin part by the National ScienceF oundationu nder grantsI ST 8416034a nd IST 8515005a nd Office of Naval Researchu nder grantN 00014-79-C-052T9h.e authorw asalso fundedb y a CorinnaB orden Keen Fellowshipw hile in residencea t Brown University.

PY - 1993/1/1

Y1 - 1993/1/1

N2 - We evaluate current explanation schemes. These are either insufficiently general, or suffer from other serious drawbacks. A domain-independent explanation theory, based on ignoring irrelevant variables in a probabilistic setting, is proposed. Independence-based maximum aposteriori probability (IB-MAP) explanations, an instance of irrelevance-based explanation, has several interesting properties, which provide for simple algorithms for computing such explanations. A best-first algorithm that generates IB-MAP explanations is presented, and evaluated empirically. The algorithm shows reasonable performance for up to medium-size problems on a set of randomly generated belief networks. An alternate algorithm, based on linear systems of inequalities, is discussed.

AB - We evaluate current explanation schemes. These are either insufficiently general, or suffer from other serious drawbacks. A domain-independent explanation theory, based on ignoring irrelevant variables in a probabilistic setting, is proposed. Independence-based maximum aposteriori probability (IB-MAP) explanations, an instance of irrelevance-based explanation, has several interesting properties, which provide for simple algorithms for computing such explanations. A best-first algorithm that generates IB-MAP explanations is presented, and evaluated empirically. The algorithm shows reasonable performance for up to medium-size problems on a set of randomly generated belief networks. An alternate algorithm, based on linear systems of inequalities, is discussed.

KW - Bayesian belief networks

KW - abduction

KW - explanation under uncertainty

KW - probabilistic reasoning

KW - relevance

UR - http://www.scopus.com/inward/record.url?scp=43949170056&partnerID=8YFLogxK

U2 - 10.1016/0888-613X(93)90027-B

DO - 10.1016/0888-613X(93)90027-B

M3 - Article

AN - SCOPUS:43949170056

VL - 8

SP - 281

EP - 324

JO - International Journal of Approximate Reasoning

JF - International Journal of Approximate Reasoning

SN - 0888-613X

IS - 4

ER -