Thepurposeof this paperis to extendverificationtests to systemswith multiple autonomousagent knowledge bases.Usinga classic approachto verification, this paperfocuseson tests concernedwithconsistency, completenessandcorrectness.In particular, this paper focusesonthoseuniqueissues that are generatedas we gofromsingle agentsystemsto multipleagentsystems. Thispaperis concernedwithinter agentverification, sincepreviousresults canbe usedfor intra agent verification. Forexample,consideroneagentwiththe rule "ifA then B"anda rule in anotheragent"irA then C".In sucha setting, the agentswouldbe constantlyat odds.Alternativelywemightfind the followingrules in oneagent("if AthenB"and"if CthenA"), while anotheragentthat interacts withthat agentmighthave the rule ("if Bthen C").Withthosetwointeractingrule basesa dialoguestarting with"A"couldcycle indefinitely. Onepotential approachto multipleagentsystemsis to comparethe knowledgebaseof eac...
Daniel E. O'Leary