BELIEF AND PROBABILITY BASED DATABASE MINING

Pawan Lingras 

Department of Computer Science, Algoma University College
Sault Ste. Marie, Ontario, Canada, P6A 2G4.
e-mail: lingras@tbird.auc.on.ca

Abstract  
This paper describes a unified approach for generating rules 
from relational databases. The relational databases have 
been successfully used in the past for generating 
probabilistic rules. The databases used for probabilistic 
rule extraction consist of tuples for which all the values 
of the attributes are precisely known. The proposed approach 
uses a belief function based generalization of the existing 
approach by allowing for a set of possible attribute values. 
Such an approach can be useful in situations where the value 
of an attribute may be unknown or may be one of a few 
possible choices. Whenever the available information is 
sufficient, the proposed method automatically generates 
probabilistic rules.

Keywords: Belief functions, expert systems, probability 
functions, rule generation, database mining.