Port a Jason agent

The Prolog incarnation keeps Jason's knowledge representation, so most of an AgentSpeak program translates line by line. The Blocks World showcase is a complete port of the Jason agent of the same name.

Beliefs and rules

parent(abraham, isaac).
ancestor(X, Y) :- parent(X, Y).

believes {
    +initialBelief { "parent"("abraham", "isaac") }
    +inferenceRule { "ancestor"(X, Y) impliedBy "parent"(X, Y) }
}

Initial goals

!start. becomes:

hasInitialGoals { !initialGoal { "start".toAtom() } }

Plans

+!introduce(P) : ancestor(A, P) <- .print(P, " descends from ", A). becomes a plan with a trigger, a guard and a Kotlin body:

prologPlan {
    adding.goal { matchingGoal { "introduce"(P) } } onlyWhen { satisfies { "ancestor"(A, P) } } triggers {
        agent.print(P, " descends from ", A)
    }
}

AgentSpeak	JaKtA
+!g	adding.​goal { matchingGoal { g } }
-!g (failure handling)	failing.​goal { matchingGoal { g } }
+b / -b	adding.​belief { matchingBelief { b } } / removing.​belief { .​.​.​ }
:​ context	onlyWhen { satisfies { context } }
[source(​S)] in a trigger	the same annotation: matchingBelief { b[source(​S)] }

Plan bodies

AgentSpeak	JaKtA
!g	agent.​achieve(​goal { g })
!!g	agent.​alsoAchieve(​goal { g })
?q	testQuery { q } (see why there are no test goals)
+b / -b	agent.​believe(​belief { b }) / agent.​forget(​belief { b })
.​print(​.​.​.​)	agent.​print(​.​.​.​)
environment actions	a skill

Messages

.send becomes a KQML function, with a MessagingSkill in scope:

AgentSpeak	JaKtA
.​send(​bob,​ tell,​ b) / .​send(​bob,​ untell,​ b)	agent.​tellTo(​bob,​ belief { b }) / agent.​untellTo(​bob,​ beliefQuery { b })
.​send(​bob,​ achieve,​ g) / .​send(​bob,​ unachieve,​ g)	agent.​delegateAchieveTo(​bob,​ goal { g }) / agent.​sendUnachieveTo(​bob,​ goal { g })
.​send(​bob,​ askOne,​ q,​ A)	agent.​askOneTo(​bob,​ beliefQuery { q })
.​send(​bob,​ askAll,​ q,​ L)	agent.​askAllTo(​bob,​ beliefQuery { q })
.​broadcast(​tell,​ b)	agent.​broadcastTell(​belief { b })

Unlike Jason, questions need an answering plan: see Ask and answer with KQML.

What changes

• Bodies are Kotlin: loops, conditionals and calls replace AgentSpeak body formulas.
• Agents live in nodes and act through skills, not a Jason environment.
• A belief has a single source: a belief told by two agents is kept once (see caveats).

See Jason-like Kotlin Agents for how JaKtA relates to Jason.