Introduction: Revising, Updating and Combining Knowledge

摘要: The question of knowledge base dynamics is currently one the most challenging problems in area intelligent information systems. A change operation, sometimes also called revision, update, consists modifying contents a body upon arrival some new piece information. What ‘a knowledge’ can take several forms ranging from database to ill-known values parameters, or set constraints incompletely describing situation. In this book, supposed refer logical sentences, uncertainty function on alternatives, each case representing beliefs an agent interest. belief literature has developed rather independently two clusters corresponding these modes representation epistemic states. oldest established revision theory based Bayes’ rule for probability measures. study been dubbed ‘probability kinematics’ [Domotor, 1985], and relevant fields such as statistics, decision philosophy science. framework emerged with advent systems artificial intelligence. field research problem naturally arises whenever evolve: how add information? what do if contradicts already stored latter cannot always be solved just by rejecting intelligence closely related issues that share lot concerns: handle defeasibility conclusions derived exception-prone knowledge? How reason about dynamic worlds? These monotonicity property classical logic inference excludes defeasibility, have close connections topics hypothetical reasoning, truth-maintenance theories action.