Argumentation

1
Argumentation
2
Argumentation

• Negotiation Mechanisms we have seen so far have
  disadvantages
• Positions cannot be justified. It cannot be
  explained how an agreement was reached. (E.g. to
  a human that ordered the agent to do something.)
• Positions cannot be changed Utility function
  does not change.
• Inflexible (Humans can change their minds during
  negotiation)
• Argumentation-based negotiation
• We will first take a semiformal look on human
  argumentation types

3
Argumentation

• Argumentation is the process of attempting to
  convince others of something
• Gilbert (1994) identified 4 modes of argument
• Logical mode If you accept that A and that A
  implies B, then you must accept that B
• Emotional mode How would you feel if it
  happened to you?
• Visceral mode Cretin!
• Kisceral mode This is against Christian
  teaching!

4
Logic based Argumentation

• Construct a proof for or against proposition of
  interest
• Formally Deduce sentence from a database ? ? f
• Well known example

human(socrates)?x human(x) ? mortal(x)
?1
?1 ? f1
f1
mortal(socrates)

• If subset G of database suffices to deduce
  sentence f we call this subset Grounds
• G ? ? G ? f

5
Logic-based Argumentation

• Basic form of logical arguments is as
  follows Database ? (Sentence, Grounds)
• where
• Database is a (possibly inconsistent) set of
  logical formulae
• Sentence is a logical formula known as the
  conclusion
• Grounds is a set of logical formulae such that
• Grounds ? Database and
• Sentence can be proved from Grounds

6
Attack and Defeat

• (f1, G1) is called non-trivial if G1 is
  consistent.
• Let (f1, G1) and (f2, G2) be arguments from some
  database DThen (f2, G2) can be defeated
  (attacked) in one of two ways
• (f1, G1) rebuts (f2, G2) if f1 ? ?f2
• (f1, G1) undercuts (f2, G2) if f1 ? ?y2 for some
  y ? G2
• A rebuttal or undercut is known as an attack

7
Acceptability Hierarchy

• Different views on the world automatically
  create contradictions between propositions in
  different agents databases ? If propositions are
  communicated ? An agents database may become
  inconsistent ? In classical logics a severe
  problem ? create hierarchy of acceptability of
  arguments
• A1 class of all arguments over ?
• A2 class of all non-trivial arguments over ?
• A3 class of all arguments over ? with no
  rebutting arguments
• A4 class of all arguments over ? with no
  undercutting arguments
• A5 class of all tautological arguments over ?
• Acceptability-Hierarchy A1(?)
  ? A2(?) ? A3(?) ? A4(?) ? A5(?)

8
Dialogue

• Our interest not argumentation with self (?
  consistency check of database) but rather
  dialogue with other agents
• Formally A step in a dialogue between two
  agents (0 and 1) is called a move. A move m
  is a pair ltagent, argumentgt
• A sequence ltm_0, m_1, m_2, ...., m_kgt is called
  dialogue history if
• agent(m_0) 0
• agent(m_i) 0 iff i is even and agent(m_i)
  1 iff i is odd. (alternating dialogue)
• agents are not allowed to make argument twice
• argument(m_i) defeats argument(m_i -1)

9
Dialogue

• Example
• m_0 (0, ltr, p, p ? q, q ? rgt)
• m_1 (1, lt(p ? q), t, t ? (p ? q)gt)
• m_2 (0, ltt, s, s ? tgt)
• ....

10
Abstract Argumentation

• Concerned with the overall structure of the
  argument (rather than internals of arguments)
• Write x ? y
• argument x attacks argument y
• x is a counterexample of y
• x is an attacker of y
• where we are not actually concerned as to what x,
  y are
• An abstract argument system is a collection or
  arguments together with a relation ? saying
  what attacks what

11
Abstract Argumentation

• Given an abstract argumentation ltX, ? gt we say
• argument x ? X is attacked by a set of arguments
  Y ? X if ?y y ? Y ? y ? x (at least one
  member of Y attacks x)
• An argument x ? X is acceptable with respect to
  a set of arguments Y ? X iff each attacker of x
  is also attacked by an argument in Y

12
An Example Abstract Argument
Ya,b,c,d,e,f,g,h,i,j,k,l,m,n,p,q
acceptable wrt. Y
not acceptable wrt. Y
13
Abstract Argumentation - Semantics

• Given an abstract argumentation ltX, ? gt we say
• A set of arguments Y is conflict free if no
  argument in Y attacks some other argument in Y
• A conflict free set of arguments Y is
• admissible iff every argument in Y is acceptable
  w.r.t. Y
• complete iff every argument acceptable w.r.t. Y
  is in Y
• grounded iff it is the smallest complete
  extension
• preferred iff it is a maximal complete extension
• stable iff every argument not in is attacked by
  an argument in Y

14
Example
A
C
D
E
B

• , A, B, C, D, E, A,D, C,E,
  B,D,E, A,D,E, ... are conflict-free
• A,B, A,C, ... are not conflict-free
• , E, A,E, A,D,E, B,D,E are admissible
• C and A,C are not admissible
• A,E is not complete
• E, A,D,E, B,D,E are complete
• E is gounded
• A,D,E and B,D,E are preferred
• A,D,E and B,D,E are stable

15
Agent Communication
16
Agent Communication

• Traditional computer sciences view on
  communication in concurrent systems (distributed
  systems)
• Solve problems like synchronizing multiple
  processes (necessary if destructive interference
  is possible)
• Example Lost update scenario Processes p1 and
  p2 shared variable v
• p1 reads v p2 reads v p2 updates v p1
  updates v
• ? updates by p2 are lost
• possible solution locking of v after read.

17
Agent Communication

• Object oriented view on communication Object o2
  invokes method m on object o1 Java o1.m(arg)
• o2 has control over invocation. o1 must invoke
  m.
• Agent view on communication Agent a2 asks agent
  a1 to perform action a. (a2 makes a request).
• a1 has control over whether it performs action
  a. Agents are autonomous.

18
Agent Communication

• What agents can do
• Perform communication acts.
• Goal Influence other agents
• E.g. to make them perform actions or to make
  them believe certain propositions
• Other agent decides whether to perform action or
  believe proposition

19
Speech Acts

• Most treatments of communication in (multi-)
  agent systems borrow their inspiration from
  speech act theory
• Speech act theories are pragmatic theories of
  language, i.e., theories of language use they
  attempt to account for how language is used by
  people every day to achieve their goals and
  intentions
• The origin of speech act theories are usually
  traced to Austins 1962 book, How to Do Things
  with Words

20
Speech Acts

• Austin noticed that some utterances are rather
  like physical actions that appear to change the
  state of the world
• Paradigm examples would be
• declaring war
• christening
• I now pronounce you man and wife -)
• But more generally, everything we utter is
  uttered with the intention of satisfying some
  goal or intention
• A theory of how utterances are used to achieve
  intentions is a speech act theory

21
Different Aspects of Speech Acts

• From A Dictionary of Philosophical Terms and
  Names
• Locutionary act the simple speech act of
  generating sounds that are linked together by
  grammatical conventions so as to say something
  meaningful. Among speakers of English, for
  example, It s raining performs the locutionary
  act of saying that it is raining, as Grablistrod
  zetagflx dapu would not.

22
Different Aspects of Speech Acts

• Illocutionary act the speech act of doing
  something else offering advice or taking a vow,
  for example in the process of uttering
  meaningful language. Thus, for example, in saying
  I will repay you this money next week, one
  typically performs the illocutionary act of
  making a promise.

23
Different Aspects of Speech Acts

• Perlocutionary act the speech act of having an
  effect on those who hear a meaningful utterance.
  By telling a ghost story late at night, for
  example, one may accomplish the cruel
  perlocutionary act of frightening a child.

24
Speech Acts

• Searle (1969) identified various different types
  of speech act
• representativesuch as informing, e.g., It is
  raining
• directiveattempts to get the hearer to do
  something e.g., please make the tea
• commissivewhich commit the speaker to doing
  something, e.g., I promise to
• expressivewhereby a speaker expresses a mental
  state, e.g., thank you!
• declarationsuch as declaring war or christening

25
Speech Acts

• There is some debate about whether this (or any!)
  typology of speech acts is appropriate
• In general, a speech act can be seen to have two
  components
• a performative verb(e.g., request, inform,
  promise, )
• propositional content(e.g., the door is
  closed)

26
Speech Acts

• Consider
• performative requestcontent the door is
  closedspeech act please close the door
• performative informcontent the door is
  closedspeech act the door is closed!
• performative inquirecontent the door is
  closedspeech act is the door closed?

27
Plan Based Semantics

• How does one define the semantics of speech acts?
  When can one say someone has uttered, e.g., a
  request or an inform?
• Cohen Perrault (1979) defined semantics of
  speech acts using the precondition-delete-add
  list formalism of planning research
• We had Plans as series of actions each action
  transforming pre-cond. into post-cond.
• Now Replace actions with speach acts Speech
  acts operate on beliefs and goals of speakers and
  listeners
• Note that a speaker cannot (generally) force a
  hearer to accept some desired mental state
• In other words, there is a separation between the
  illocutionary act and the perlocutionary act

28
Plan-Based Semantics

• Here is their semantics for requestrequest(s,
  h, f)
• pre
• s believe h can do f(you dont ask someone to do
  something unless you think they can do it)
• s believe h believe h can do f(you dont ask
  someone unless they believe they can do it)
• s believe s want f(you dont ask someone unless
  you want it!)
• post
• h believe s believe s want f(the effect is to
  make them aware of your desire)

29
Agent Communication Languages

• DARPA funded KSE (Knowledge Sharing Effort) in
  early 1990s designed two agent communication
  languages (ACLs) with different purpose
• KQML (Knowledge Query and Manipulation
  Language).
• KIF (Knowledge Interchange Format).

30
KQML and KIF

• KQML is an outer language, that defines various
  acceptable communicative verbs, or
  performativesExample performatives
• ask-if (is it true that. . . )
• perform (please perform the following action. .
  . )
• tell (it is true that. . . )
• reply (the answer is . . . )
• KIF is a language for expressing message content

31
KIF

• First order logic in LISP notation
• Contains most constructs from FOL
• VERY expressive (too expressive ?)
• Possible to express e.g.
• properties of things in a domain (Carla has red
  hair) (George has a silly little beard)
• relationships between things in a domain (John
  is married to Eva)
• general properties of a domain (everybody has a
  mother)
• Quantifiers forall exists
• Standard vocabulary (Ontology) for objects
  (numbers, strings etc., complex objects, lists
  etc.) and methods on them

32
KIF Knowledge Interchange Format

• The temperature of m1 is 83 Celsius(
  (temperature m1) (scalar 83 Celsius))
• An object is a bachelor if the object is a man
  and is not married(defrelation bachelor (?x)
  (and (man ?x) (not (married ?x))))
• Any individual with the property of being a
  person also has the property of being a
  mammal(defrelation person (?x) gt (mammal ?x))

33
KQML - Knowledge Query and Manipulation Language

• Each message has performative (class of
  message) and a number of parameters

(ask-one content (PRICE IBM ?PRICE) receiver
stockServer language LPROLOG ontology
NYSE-TICKS)
34
KQML
Parameter Meaning
content of message formal language message is
interminology message is based onwill sender
ever deny content of message?reply expected?
identifier of reply?id of replysenderreceiver
content language ontology
force reply-with in-reply-to sender rec
eiver
35
KQML and KIF

• In order to be able to communicate, agents must
  have agreed on a common set of terms
• A formal specification of a set of terms is known
  as an ontology
• The knowledge sharing effort has associated with
  it a large effort at defining common ontologies
  software tools like ontolingua for this purpose
• Example KQML/KIF dialogueA to B (ask-if (gt
  (size chip1) (size chip2)))B to A (reply
  true)B to A (inform ( (size chip1) 20))B to
  A (inform ( (size chip2) 18))

36
FIPA

• More recently, the Foundation for Intelligent
  Physical Agents (FIPA) started work on a program
  of agent standards the centerpiece is an ACL
• Basic structure is quite similar to KQML
• performative20 performative in FIPA
• housekeepinge.g., sender, etc.
• contentthe actual content of the message

37
FIPA

• Example(inform sender agent1 receiver agent5
  content (price good200 150) language sl ont
  ology hpl-auction)

38
FIPA
39
FIPA ACL Performatives
accept-proposal
sender accepts proposal made by other agent
agree
sender agrees to carry out requested action
cancel
follows request indicates intention behind
request is not valid any more
call for proposals initiates negotiation between
agents content-parameter contains action
(desired to be done by some other agent) (e.g.
sell me car) and condition (e.g. price lt
1000)
cfp
confirm
confirm truth of content (recipient was unsure)
disconfirm
confirm falsity of content (recipient was unsure)
attempt to do requested action failed
failure
40
FIPA ACL Performatives
inform
together with request most important
performative basic mechanism for communicating
information sender wants recipient to believe
info sender believes info itself
inform-if
informs other agent about truth of statement (in
its content parameter) if it is true typically
content of request message (thus asking the
receiver to inform me if statement is true)
inform-ref
informs other agent about value of expression (in
its content parameter) typically content of
request message (thus asking the receiver to give
me value of expression)
41
FIPA ACL Performatives
not-understood
sender indicates that it recognized that an
action was performed by other agent but it did
not understood why it was performed. (? error
handling mechanism)
propagate
request to propagate a message to specified agents
propose
make proposal
same as propagate but with proxy functionality
proxy
direct query for the truth of a statement
query-if
direct query for the value of an expression
query-ref
reject request
refuse
reject-proposal
sender does not accept proposal
42
FIPA ACL Performatives
issue request for an action
request
issue request to do action if and when a
statement is true
request-when
request-whenever
issue request to do action if and whenever a
statement is true
subscribe
sender asks to be notified when statement changes
43
Inform and Request

• Inform and Request are the two basic
  performatives in FIPA. All others are macro
  definitions, defined in terms of these.
• The meaning of inform and request is defined in
  two parts
• pre-conditionwhat must be true in order for the
  speech act to succeed
• rational effectwhat the sender of the message
  hopes to bring about

44
Inform and Request

• For the inform performativeThe content is a
  statement.Pre-condition is that sender
• holds that the content is true
• intends that the recipient believe the content
• does not already believe that the recipient is
  aware of whether content is true or not

45
Inform and Request

• For the request performativeThe content is an
  action.Pre-condition is that sender
• intends action content to be performed
• believes recipient is capable of performing this
  action
• does not believe that receiver already intends to
  perform action

46
Formal Semantics

• Semantics of FIPA-ACL is formally specified in
  language SL. Based on Cohen Levesques theory
  of speech acts as rational action.
• SL allows to represent beliefs, desires,
  uncertain beliefs.

47
Formal Semantics

• Definitions of request and inform.
  Sub-Expressions
• Bif agent i beliefs f
• Bi fif agent i has definite opinion concerning
  truth of f
• Ui fif agent i is uncertain about f
• agent(a,j) agent that can do a is j
• Ij(a) agent j intends to do a
• Done(a) action a has been done

lti, inform(j,f)gt
feasibility precond
Bjf
rational effect
lti, request(j,a)gt
feasibility precond
Done(a)
rational effect