## Reasoning about Beliefs: An Inference Network Approach (1994)

Citations: | 1 - 0 self |

### BibTeX

@MISC{Low94reasoningabout,

author = {Boon Toh Low},

title = {Reasoning about Beliefs: An Inference Network Approach},

year = {1994}

}

### OpenURL

### Abstract

This thesis studies a hybrid symbolic-neural network for representing and reasoning about commonsense beliefs called a Neural-Logic Belief Network (NLBN). Propositional concepts are represented by nodes and relations amongst the concepts are represented by numerical directed links. NLBN employs neural network-like computation functions and it can be interpreted by a competitive voting semantic. Classical relations such as AND, OR and NOT, as well as human biased relations can be modeled by the numerical links. To represent incomplete beliefs, this belief network uses a four-valued scheme (believe the concept, believe its negation, unknown, and contradictory) but two-valued reasoning can be achieved using Closed World Default reasoning on selected propositions. The different reliabilities of beliefs are expressed as linguistical degrees in a total asymmetric order. Unlike the material implication in logic systems, defeasible IF-THEN rules and inheritance relations in this formalism are...