Causes for events: Their computation and applications

Abstract

A formal definition is presented for the cause for an event with a given knowledge base. An event is understood to be any closed formula; a knowledge base is a set of closed formulae; and a cause is a prenex formula, the matrix of which is a conjunction of literals. The properties of minimality, basicness, consistency and nontriviality are defined to characterise causes that are useful and interesting. An algorithm for computing basic, nontrivial and minimal causes is presented, and its soundness is proved. An extensive example is provided to illustrate the application of causes. In this example is discussed the interaction between the knowledge base, event and causes, and in particular, the property of consistency is used to eliminate ambiguity, and to discover deficiencies of the knowledge base.