44
1 Computer Science Department California Polytechnic State University San Luis Obispo, CA, U.S.A. Franz J. Kurfess CPE/CSC 580: Intelligent Agents 1

1 Computer Science Department California Polytechnic State University San Luis Obispo, CA, U.S.A. Franz J. Kurfess CPE/CSC 580: Intelligent Agents 1

Embed Size (px)

Citation preview

1

Computer Science DepartmentCalifornia Polytechnic State University

San Luis Obispo, CA, U.S.A.

Franz J. Kurfess

CPE/CSC 580: Intelligent Agents

1

3© Franz J. Kurfess

Course Overview❖ Introduction

Intelligent Agent, Multi-Agent Systems Agent Examples

❖ Agent Architectures Agent Hierarchy, Agent Design Principles

❖ Reasoning Agents Knowledge, Reasoning, Planning

❖ Learning Agents Observation, Analysis, Performance Improvement

❖ Multi-Agent Interactions Agent Encounters, Resource Sharing, Agreements

❖ Communication Speech Acts, Agent Communication Languages

❖ Collaboration Distributed Problem Solving, Task and Result Sharing

❖ Agent Applications Information Gathering, Workflow, Human Interaction, E-Commerce, Embodied Agents, Virtual Environments

❖ Conclusions and Outlook

4© Franz J. Kurfess

Overview Communication among Agents

❖ Motivation

❖ Objectives

❖ Communication speech acts; agent communication languages

❖ Cooperation self-interest, societal benefits

❖ Important Concepts and Terms

❖ Chapter Summary

7© Franz J. Kurfess

Motivation

8© Franz J. Kurfess

Objectives

10

Communication

Communication BasicsSpeech Acts

Language: Syntax, Semantics, Pragmatics

11

© Franz J. Kurfess

Basic Concepts

❖ communication exchange of information requires a shared system of signs greatly enhanced by language speaker

produces signs as utterances general: not only spoken language

listener (hearer) perceives and interprets signs

12

[Woolridge 2009]

Communication among Agents

13

© Franz J. Kurfess

Purpose of Communication

❖ sharing of information among agents or systems query other agents for information responses to queries requests or commands

actions to be performed for another agent offer

proposition for collaboration acknowledgement

confirmation of requests, offers sharing

of experiences, feelings

14

© Franz J. Kurfess

Communication Problems

❖ intention what is the expected outcome (speaker’s perspective)

❖ timing when is a communication act appropriate

❖ selection which act is the right one

❖ language what sign system should be used

❖ interpretation will the intended meaning be conveyed to the listener

❖ ambiguity can the intention be expressed without the possibility of misunderstandings

15

© Franz J. Kurfess

Language and Communication

❖ Natural Language used by humans evolves over time moderately to highly ambiguous

❖ Formal Languages invented rigidly defined little ambiguity

16

© Franz J. Kurfess

Natural Language

❖ formal description is very difficult sometimes non-systematic, inconsistent, ambiguous

❖ mostly used for human communication easy on humans tough on computers

❖ context is critical situation, beliefs, goals

17

© Franz J. Kurfess

Formal Languages

❖ symbols terminal symbols

finite set of basic words not: alphabet, characters

non-terminal symbols intermediate structures composed of terminal or non-

terminal symbols

❖ strings sequences of symbols

❖ phrases sub-strings grouping important parts of a string

18

© Franz J. Kurfess

Formal Languages Cont.

❖ sentences allowable strings in a language composed from phrases

❖ grammar rules describing correct sentences often captured as rewrite rules in BNF notation

❖ lexicon list of allowable vocabulary words

19

© Franz J. Kurfess

Communication Models

❖ encoded message model a definite proposition of the speaker is encoded into

signs which are transmitted to the listener the listener tries to decode the signs to retrieve the

original proposition errors are consequences of transmission problems

❖ situated language model the intended meaning of a message depends on the

signals as well as the situation in which they are exchanged

mis-interpretation may lead to additional problems

20

© Franz J. Kurfess

Communication Types

❖ telepathic communication speaker and listener have a shared internal

representation communication through Tell/Ask directives

❖ language-based communication speaker performs actions that produce signs which other

agents can perceive and interpret communication language is different from the internal

representation more complex

involves several mappings language needs to be generated, encoded, transmitted, decoded,

and interpreted

21

© Franz J. Kurfess

Telepathic Communication

[Russell & Norvig 1995]

22

© Franz J. Kurfess

Language-Based Communication

[Russell & Norvig 1995]

23

© Franz J. Kurfess

Communication Steps: Speaker

❖ intention decision about producing a speech act

❖ generation conversion of the information to be transferred into the

chosen language

❖ synthesis actions that produce the generated signs

24

© Franz J. Kurfess

Communication Steps: Listener

❖ perception reception of the signs produced by the speaker

speech recognition, lip reading, character recognition analysis

syntactic interpretation (parsing) semantic interpretation

disambiguation selection of the most probable intended meaning

incorporation the selected interpretation is added to the existing world

model as additional piece of evidence

25

© Franz J. Kurfess

Communication Example

[Russell & Norvig 1995]25

26

Speech Acts

BasicsSpeech Act Theory

MappingsComponentsSemantics

27

© Franz J. Kurfess

Speech Act

❖ used for the production of language

❖ independent of the communication mode talking, sign language, typing, flags

❖ word basic meaningful communicative sign smaller entities may exist

e.g. syllable, phonem, letter don’t carry meaning

❖ speaker (sender) producer of an utterance

❖ hearer (listener, recipient) consumer of an utterance

28

© Franz J. Kurfess

Speech Act Theory

❖ developed in linguistics, cognitive science, communication theory

❖ pragmatic theories of language based on language use

❖ utterances elementary speech actions based on or related to intentions

❖ different typologies of speech acts

29

[Woolridge 2009] 8-8

Speech Acts - Searle Searle (1969) identified various different types of

speech act: representatives:

such as informing, e.g., ‘It is raining’ directives:

attempts to get the hearer to do something e.g., ‘please make the tea’

commissives:which commit the speaker to doing something, e.g., ‘I promise to… ’

expressives:whereby a speaker expresses a mental state, e.g., ‘thank you!’

declarations:such as declaring war or christening

30

[Woolridge 2009] 8-9

Speech Act Components

a performative verb: e.g., request, inform, promise, …

propositional content: e.g., “the door is closed”

31

[Woolridge 2009] 8-10

Speech Act Mappings Speech act <=> performatives & content:

performative = requestcontent = “the door is closed”speech act = “please close the door”

performative = informcontent = “the door is closed”speech act = “the door is closed!”

performative = inquirecontent = “the door is closed”speech act = “is the door closed?”

32

[Woolridge 2009]

Speech Act Semantics

intention of the speaker leads to a specific formulation of a statement

interpretation by the listener may be different from the intended meaning

methods from other AI areas have been applied e.g. planning

33

Agent Communication

Languages

standard formats for the exchange of knowledge and information

usually based on messages

34

[Woolridge 2009]

KQML KQML (Knowledge Query and Manipulation

Language) developed by the ARPA knowledge sharing

initiative KIF (Knowledge Interchange Format)

designed to work in conjunction with KQML

35

[Woolridge 2009]

KQML and KIF KQML is an ‘outer’ language

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 related to knowledge representation languages

36

[Woolridge 2009]

KIF – Knowledge Interchange Format Used to state: Properties of things in a domain

e.g., “Orna is chairman” Relationships between things in a domain

e.g., “Michael is Yael’s boss” General properties of a domain

e.g., “All students are registered for at least one course”

37

[Woolridge 2009]

KIF Examples “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) :=> (mammal ?x))

38

[Woolridge 2009]

KQML and KIF communication between agents requires a common

set of terms ontology

formal specification of a set of terms knowledge sharing

requires defining common ontologies OWL - Web Ontology Language ontology editors

Protégé

39

[Woolridge 2009]

KQML/KIF Dialogue Example

A to B: (ask-if (> (size chip1) (size chip2)))

B to A: (reply true)

B to A: (inform (= (size chip1) 20))

B to A: (inform (= (size chip2) 18))

40

[Woolridge 2009]

Criticisms of KQML fluid performative set

leading to interoperability problems transport mechanisms not precisely defined semantics not rigorously defined missing commissives

performatives for making commitments performative set too large and ad hoc

41

[Woolridge 2009]

FIPA Agent Communication Language program of agent standards

initiated by the Foundation for Intelligent Physical Agents (FIPA)

the centerpiece is an ACL structure similar to KQML

performatives20 performatives in FIPA

contentthe actual content of the message

housekeepinge.g., sender, receiver, ...

42

[Woolridge 2009]

FIPA ACL Example

Example:(inform

:sender agent1:receiver agent5:content (price good200

150):language sl:ontology hpl-auction

)

43

[Woolridge 2009]

FIPA Performatives

44

[Woolridge 2009]

“Inform” and “Request”

two basic performatives in FIPA all others are macro definitions defined in terms of “Inform” and “Request”.

semantics of “Inform” and “Request” pre-condition

what must be true in order for the speech act to succeed “rational effect”

what the sender of the message hopes to bring about

45

[Woolridge 2009]

“Inform”

pre-condition is that the 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 content is a statement

46

[Woolridge 2009] 8-24

“Request”

pre-condition is that the 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 content is an action

49

© Franz J. Kurfess

Summary Communication

50

© Franz J. Kurfess

Important Concepts and Terms

❖ agent

❖ Agent Communication Language

❖ alphabet

❖ ambiguity

❖ communication

❖ collaboration

❖ coordination

❖ formal language

❖ grammar

❖ hearer

❖ KIF

❖ KQML

❖ language

❖ lexicon

❖ listener

❖ multi-agent system

❖ natural language

❖ pragmatics

❖ recipient

❖ semantics

❖ sender

❖ sign

❖ speech act

❖ syntax

❖ utterance

❖ vocabulary