UNIT-III

Dr. Shivangi Singh

MISSION VISION CORE VALUES

CHRIST is a nurturing ground for an individual’s Excellence and Service Faith in God | Moral Uprightness
holistic development to make effective contribution to Love of Fellow Beings
the society in a dynamic environment Social Responsibility | Pursuit of Excellence
 CHRIST
Deemed to be University

Contents

● Logical & Physical Distribution

● Traditional Distributed systems Vs Multiagent Communication Concept

Excellence and Service

 CHRIST
Deemed to be University

Limitation of traditional protocol specifications (like Finite

State Machines - FSMs) for Multi-Agent Systems (MAS)
● limitation of traditional protocol specifications (like Finite State Machines -
FSMs) for Multi-Agent Systems (MAS) compared to traditional distributed
systems.

Buyer Seller

Excellence and Service

 CHRIST
Deemed to be University

● The figure describes a common way to specify interaction protocols

(like the purchase protocol example), often visualized as a Finite State
Machine (FSM).

Focus: These protocols define:

● Message Schemas: The structure/data format of messages (e.g.,

Offer(book, price), Accept, Reject, Update Offer (new Price)).

● Legal Flows: The allowed sequence and conditions for sending

messages (e.g., Seller sends Offer first; Buyer can then Accept or
Reject; If Buyer Accepts, Seller can send Update Offer).

Excellence and Service

 CHRIST
Deemed to be University
Core Limitation :

● It lacks explicit public semantics (the real-world meaning, commitments, and effects of
messages).

● Traditional protocol specs (like the FSM in Buyer seller example) only define syntax and
flow, not meaning.

● They do not specify the real-world, social, or commitment-based semantics of the

messages exchanged between agents.

● Example: The message "Offer" in the protocol. What does it mean when an agent sends it?
○ Does it simply convey information ("I propose X")?

○ Or does it create a social commitment (e.g., "I, the Seller, am now publicly committed
to selling you this book at $Y if you accept it before I update or withdraw")?

○ What does "Update Offer" semantically do? Does it cancel the old commitment and
create a new one, or merely modify the existing one?
Excellence and Service
 CHRIST
Deemed to be University

Why is this Limitation a Problem for MAS (Heterogeneity &

Interoperability)?

● Heterogeneity: MAS involve diverse agents, potentially built by different

developers/organizations with different internal architectures and interpretations.
● Semantic Mismatch: If the protocol only specifies flow/syntax, agents can interpret
the meaning of messages differently.
○ Concrete Problem: Buyer might interpret "Offer" as creating a binding social
commitment from the Seller. Seller might interpret "Offer" merely as a
non-binding suggestion. Their internal models of the interaction are misaligned.

● Breakdown Risk: This semantic mismatch can lead to critical misunderstandings

and interaction failures.
○ Example: Buyer accepts an offer, believing the Seller is committed. Seller then
sends an Update Offer (increasing price), believing the original offer was never
binding. Buyer sees this as the Seller breaking a commitment. Trust breaks down,
the interaction fails, or unexpected disputes arise.

Excellence and Service

 CHRIST
Deemed to be University

● Non-Interoperability: Agents built with conflicting interpretations of

the message semantics cannot reliably and correctly interact, even if
they perfectly follow the syntactic protocol flow. True
interoperability requires shared understanding of meaning.

The Proposed Solution (Accommodating Semantic Heterogeneity):

○ To make MAS protocols robust against semantic heterogeneity, the
specification must make the meanings of the messages public and
explicit.
○ This involves defining the social commitments, obligations,
permissions, and intentions created, modified, or discharged by each
message within the interaction context. (The text hints at "social
commitments" as a key concept for this).

Excellence and Service

 CHRIST
The Paradox of Practice: Deemed to be University

The text acknowledges that many MAS protocols are specified only
as flows (syntax + ordering) and often work "fairly well".
This might be because:
a. The domain is simple or well-understood (semantics are implicit).
b. Agents are built by the same team/organization (shared implicit understanding).
c. Agents are simple/reactive and don't deeply reason about commitments.
d. Failure modes are tolerated or handled outside the protocol.
However, the fig. argues this is fundamentally inadequate for achieving true
interoperability and robustness in open, heterogeneous MAS environments where
deep semantic alignment is crucial.

Excellence and Service

 CHRIST
Deemed to be University

Why do we need Agent Communication?

a. Multi agent systems allow distributed problem solving

b. This requires the agents to coordinate their actions
c. Agent communication facilitates this by allowing
individual agents to interact allows cooperation
allows information sharing

Excellence and Service

 CHRIST
Deemed to be University

Communication Categories

We can classify communication in a mechanistic manner

a. via the type of Sandee-addressee link
b. via the nature of the medium or
in a higher-level meaning-based manner
a. via the type of intention

Excellence and Service

 CHRIST
Deemed to be University

The Sendee - Addressee Link

Communication can be

Point to Point An agent talks directly to another agent

Broadcast :- An agent sends some information to a group of agents

Mediated The communication between two agents is mediated by a

third party
Example: facilitators

Excellence and Service

 CHRIST
Deemed to be University

Excellence and Service

 CHRIST
Deemed to be University

Excellence and Service

 CHRIST
Deemed to be University

Excellence and Service

 CHRIST
Deemed to be University

Agents Autonomy in Communication

● Agents choose when/what/how to communicate (e.g., refuse requests,

ignore messages).
● Challenge: Balancing individual autonomy with system-wide coordination.

Concept of Logical & Physical Distribution

Physical Distribution
● Definition: Agents run on separate hardware/network nodes.
● Impact on Communication:
○ Challenges: Network latency, bandwidth, fault tolerance.
○ Solutions: Message queues (e.g., RabbitMQ), TCP/IP protocols.
● Example: IoT devices in a smart factory.

Excellence and Service

 CHRIST
Deemed to be University

Physical Distribution

Agent A Agent B Agent C

(Loc1) (Loc2) (Loc3)

Network (Internet)
Latency, Bandwidth, Failures

Key Characteristics:

Communication: Network-dependent (TCP/IP, MQTT, Bluetooth).

Failure Resilience: Fault-tolerant (one node fails; others survive).

Autonomy Impact: Agents physically independent; may act despite network partitions.

Examples:

Drone swarms in disaster response.

Excellence and Service

 CHRIST
Deemed to be University
Physical Communication Concept
Definition:
Physical communication refers to the concrete mechanisms that enable data transfer between agents —
focusing on how the data physically moves across a network or medium.
Key Elements:
•Networking Infrastructure: LAN, WAN, ad-hoc networks for robots, wireless sensor networks.
•Transport Protocols: TCP/IP, UDP, or specialized middleware like JADE, ROS, or MQTT.
•Bandwidth & Latency: Physical constraints that can impact communication reliability and timeliness.
•Connectivity Topologies: Centralized broker, peer-to-peer, mesh networks, etc.
•Security Mechanisms: Encryption, authentication, and fault-tolerance to ensure safe transfer.
Example:
A swarm of drones shares position data via a Wi-Fi mesh network using ROS messages over UDP.
Here, packet loss, range, and interference are physical concerns.

Excellence and Service

 CHRIST
Logical Distribution Deemed to be University

● Definition: Agents are software entities on the same physical system but
operate independently.
● Impact on Communication:
○ Challenges: Shared resources, concurrency, coordination overhead.
○ Solutions: Thread-safe message passing (e.g., gRPC, shared memory, Inter process
communication).
● Example: ChatGPT plugins (agents) collaborating on one server;
Microservices on one server.

Agent A Agent B Agent C

(Planning) (Sensing) (Acting)

IPC IPC
Shared Memories ,RPC ,Queues

Autonomy Impact: Agents control when/how to interact but share resources.

Excellence and Service
 CHRIST
Deemed to be University
Key Elements of Logical Communication
Protocols: The predefined rules or interaction patterns that define the sequence and
semantics of message exchange (e.g., Contract Net Protocol).
Ontologies: Shared vocabularies that ensure agents interpret message content
consistently.
Speech Acts: Based on theories like KQML or FIPA ACL, messages represent
intentions (e.g., inform, request, query).
Dialogue Policies: Higher-level structures governing multi-turn interactions and
negotiations.
Example: Two agents negotiating task allocation. One sends a cfp (Call for Proposal);
the other responds with a propose

Excellence and Service

 CHRIST
Deemed to be University

Conceptual Foundations of Communications in MAS

Excellence and Service

 CHRIST
Deemed to be University

Excellence and Service

 CHRIST
Deemed to be University

Excellence and Service

 CHRIST
Deemed to be University

Conceptual Foundations of Communications in MAS

● Communicative Acts
the study of communication is speech act theory, better known as
communicative act theory.

Same content of the message, different meanings (depending on the used performative)

Excellence and Service

 CHRIST
Deemed to be University

Communication Standards

● Agents must understand each other even if running on different machines

and/or different operating systems
Standards
● allow different groups to write cooperating agents
● help abstract out communication, by defining high-level general languages
and protocols

Excellence and Service

 CHRIST
Deemed to be University

● Communication Protocols and Data Formats:

○ MAS to be handled by agents who will utilize different communication mechanisms (e. g. g. ,

, and are amongst the protocols while JSON, BSON, and Avro are used for data formatting.

Datasets may be held either in a simple presentation format (e. g. CSV, JSON, XML, and

Protocol Buffers) by their core technology and architecture.

○ Language incompatibility and the use of various formats and protocols can cause the loss of

data, communication failure and reduce the effectiveness of information sharing.

Excellence and Service

 CHRIST
Deemed to be University

Types of communicative acts:

Informing
An agent shares information with other agents.

It provide data, knowledge, or state updates without expecting an immediate response.

Ex:
In smart grid system the energy sensor agent informs the control center agent about
current power consumption levels.

Requesting:
An agent request another agent to perform a some action.

The requesting agent expects some form of response, either an acknowledgment or a

result of the requested action.

Excellence and Service

 CHRIST
Deemed to be University

● Ex: The control center agent requests a storage agent to discharge stored
energy to balance the grid.

Agreeing
● An agent agreeing the proposal or request made by another agent.

● This act confirms that the agent will comply with the proposed action or plan.

Ex: The storage agent agrees to the request and begins discharging energy.

Refusing
● An agent declines a request or proposal from another agent.

● This act communicates its inability or unwillingness to perform the requested

action.

● Ex: A generator agent refuses to increase output due to maintenance issues.

Excellence and Service
 CHRIST
Deemed to be University

Proposing
● Agent suggests a plan, action, or solution to other agents.

● This act invites feedback, negotiation, or acceptance from the recipients.

Ex: A renewable energy agent proposes a schedule for optimal energy

production based on weather forecasts.

Querying
● An agent asks for specific information from another agent.

● This act involves looking for data, knowledge, or status updates and
expecting an informative response.

Ex: The control center agent queries all generation agents for their current
status.
Excellence and Service
 CHRIST
Deemed to be University

Promising
● An agent pledges to performing a specific action in the future.

● This act involves making a guarantee that the agent will follow through with
the stated commitment.

● Ex: A generator agent promises to come online within two hours.

Apologizing
● An agent expresses regret or acknowledges a mistake or failure.

● This act helps maintain trust and smooth relationships between agents.

● An agent apologizes for the delayed response due to network latency.

Excellence and Service

 CHRIST
Deemed to be University

Negotiating
● Agents engage in a back-and-forth exchange to reach a mutually acceptable
agreement.

● This involves making proposals, counter-proposals, concessions, and

agreements.

Ex: Agents negotiate the allocation of energy resources to meet demand

efficiently

Warning
● An agent alerts others to a potential issue or threat. This act is proactive and
aims to prevent problems or risks.

Ex: A monitoring agent warns the system about an unusual spike in power
consumption.
Excellence and Service
 CHRIST
Deemed to be University

Agent Communication Primitives

It is a fundamental operations or actions that agents use to interact and

communicate with each other in a multi-agent system (MAS).

These primitives form the basic building blocks for higher-level communication
protocols and interaction patterns

Ex: Send, receive, cancel

In the stock-trading domain, we would see primitives such as request stock

quote and provide stock quote.

In the electronic commerce domain, we would see primitives such

as quote price, quote delivery charges, and so on.

The semantics of the primitives would be expressed in commitments.

Excellence and Service
 CHRIST
Deemed to be University

● Similar primitives may end up with completely different meanings, reflecting

the needs and practices of the applicable domains.

● For example, in typical practice, a price quote is an offer to sell, meaning that
the seller becomes committed to providing the specified item at the quoted
price.

● In contrast, in typical practice, a stock quote carries no such implication of an

offer to sell

● The quoted price is the price at which the previous transaction was
completed on the specified stock symbol,

● Not that the brokerage who provided the quote is offering to sell you the stock
for the quoted price.

Excellence and Service

 CHRIST
Deemed to be University

● Clearly, a protocol that specifies only the flow of messages, such as the one
in Figure 3.1, does not capture such subtleties of meaning.

Criteria for Evaluation

● Communication has been studied in software engineering, distributed

systems, and distributed artificial intelligence.

● Consequently, there are many approaches for specifying protocols

● Criteria to evaluate each approach

In Software engineering
Protocol specifications should not be far removed from the expression of
stakeholder requirements.

Excellence and Service

 CHRIST
Deemed to be University

Common Challenges in MAS

● Heterogeneity: Diverse agents may use different protocols or networks.
● Scalability: Maintaining logical coherence as the number of agents grows.
● Fault Tolerance: Handling network failures while preserving logical
agreements.
● Dynamic Topologies: Agents may join or leave dynamically (e.g., in ad-hoc
robotic swarms).
Practical Frameworks
● JADE (Java Agent Development): Implements FIPA standards, provides
logical ACL and physical messaging over RMI/TCP.
● ROS (Robot Operating System): Abstracts message passing via topics and
services, with physical transport over TCPROS/UDPROS.
● MQTT/ZeroMQ: Used for lightweight publish-subscribe models in distributed
sensor-agent networks.

Excellence and Service

 CHRIST
Deemed to be University

Excellence and Service

 CHRIST
Deemed to be University

Excellence and Service

 CHRIST
Deemed to be University

Excellence and Service

 CHRIST
Deemed to be University

Excellence and Service

 CHRIST

Commitment based multi agent approaches

Deemed to be University

● Commitment-based multiagent approaches prioritize the business meanings

of service engagements, captured through the participants’ commitments to
one another. Each participant is modeled as an agent, and service
engagements are carried out by creating and manipulating commitments to
one another.
● A commitment is an expression of the form
C(debtor, creditor, antecedent, consequent) ,
● where debtor and creditor are agents, and antecedent and consequent are
propositions.
● A commitment C(x,y, r,u) means that x is committed to y that if r holds,
then it will bring about u. If r holds, then C(x, y, r,u) is detached, and the
commitment
● C(x,y,r,u) holds (r being the constant for truth). If u holds, then the
● commitment is discharged and does not hold any longer. All commitments
are conditional; an unconditional commitment is merely a special case where
the antecedent equals
Excellence and Service
 CHRIST
Deemed to be University

Commitments can be manipulated to support flexibility using

operations like CREATE, CANCEL, RELEASE, DELEGATE, and ASSIGN:
•CREATE(x, y, r, u): performed by x, causes C(x, y, r, u) to hold.
•CANCEL(x, y, r, u): performed by x, causes C(x, y, r, u) to not hold.
•RELEASE(x, y, r, u): performed by y, causes C(x, y, r, u) to not hold.
•DELEGATE(x, y, z, r, u): performed by x, causes C(z, y, r, u) to hold.
•ASSIGN(x, y, z, r, u): performed by y, causes C(x, z, r, u) to hold.
•DECLARE(x, y, r): performed by x to inform y that r holds (may indirectly
affect commitments by causing detaches and discharges).
Commitments arise in a social or legal context, which defines the rules
of encounter and may serve as an arbiter, imposing penalties for
violations.

Excellence and Service

 CHRIST
Deemed to be University

● Further, a commitment arises in a social or legal context.

● The context defines the rules of encounter among the interacting parties, and
often serves as an arbiter in disputes and imposes penalties on parties that
violate their commitments.

Commitment Protocol Specification

Table 3.1 shows the specification of a commitment protocol between a merchant
and a customer

Excellence and Service

 CHRIST
Deemed to be University

Distinguishing message syntax and meaning: two views of the same enactment.

Excellence and Service

 CHRIST
Deemed to be University
Evaluation with Respect to MAS

Compliance: Protocol enactments can be judged correct as long as the parties

involved do not violate their commitments.

A customer would be in violation if it keeps the goods but fails to pay.

In this manner, commitments support business-level compliance and do not

dictate specific operationalizations.

Flexibility : Enhances flexibility over traditional approaches by expanding the

operational choices for each party.

For example, if the customer substitutes a new way to make a payment or elects
to pay first.

Customer may return damaged goods for credit; and so on

Excellence and Service
 CHRIST
Software Engineering Deemed to be University

Commitments offer a high-level abstraction for capturing business interactions

Commitment-based approaches offer a compelling alternative to the traditional

SE approaches for building systems comprised of autonomous agents.

Abstraction –reducing complexity

internal mental states and processes of intelligent agents

Lexical basis- Assigning meaningful names

Excellence and Service
 CHRIST
Deemed to be University

● In a Multi-Agent System (MAS), commitment-based protocols:

● ✅ Ensure compliance by checking if obligations are fulfilled.
● ✅ Allow flexibility by not prescribing exact steps — agents decide how to
meet commitments.

Excellence and Service

 CHRIST
Deemed to be University

Disaster Response Workflow

● Emergency declared with location/supply needs
● Available drones identified and verified
● Depot status confirmed via header check
● Drone capabilities interrogated
● Weather monitoring subscription established
● Mission control delegated to another team
● Flight path updated due to obstacles
● Sensor data requested in specific format
● Delivery notification pushed automatically
● Damaged drone decommissioned
● Weather alert interrupts operations
Excellence(async)
and Service
 CHRIST
Deemed to be University
Apply the concept of communication acts on drones acting in disaster relief
system
Communicative Act HTTP Method & Endpoint Description

Declaration POST /emergencies Announcing new emergency situation

Acquisition GET /drones Retrieving available drone status

Directive PATCH /missions/{id}/path Modifying drone flight path

Revocation DELETE /drones/{id} Decommissioning damaged drone

Subscription POST /subscriptions Registering for weather alerts

Verification HEAD /supply-depots/{id} Checking depot status (headers only)

GET /drones/{id}/sensor-data+
Negotiation Requesting specific data format
Accept header
Delegation POST /missions/{id}/delegate Transferring mission control
Interrogation OPTIONS /drones/{id} Querying drone capabilities
Push alerts for delivery completion and
Notification Webhooks
weather changes

Lets see the work on Disaster Response Flow

Excellence and Service

 CHRIST
Deemed to be University
REST API Design Patterns Explained Using a Drone Relief Scenario

1. Resource-Based URLs
● Identify core resources in drone operations:
● drones
● missions
● relief packages
● emergency zones

HTTP Method Endpoint Description

GET /drones List all registered drones
POST /missions Create a new relief mission
Retrieve details of a specific
GET /drones/{drone-id}
drone
PUT /missions/{mission-id} Update mission coordinates
GET /emergency_zones List active disaster zones

Excellence and Service

 CHRIST
Deemed to be University
HTTP Methods = Actions

HTTP Method Endpoint Action Description

POST /missions Create Launch a new mission
GET /missions/{id} Read Check mission status
PUT /missions/{id} Update Reroute a mission
DELETE /missions/{id} Delete Abort a mission

This sequence shows how all communicative acts interact during a crisis:

● Synchronous acts (1-9) follow request-response patterns

● Asynchronous acts (10) push critical updates immediately

● Verification/Interrogation provide quick status checks

● Delegation/Subscription enable dynamic team coordination

● Revocation/Directive ensure operational safety

● The system maintains full auditability with each act generating timestamped logs - crucial for
Excellence and Service
post-disaster analysis and improving future relief operations.
 CHRIST
Deemed to be University

Excellence and Service

 CHRIST
Deemed to be University

Agent Primary Responsibility Communicative Acts

Field Team Crisis decision-making Declaration, Delegation, Directive

API Orchestrator System coordination Negotiation, Interrogation, Verification

Drone Fleet Physical operations Notification, Directive response

Supply Depot Resource management Verification response
Weather System Environmental monitoring Notification (push)

Excellence and Service

Agent Communication

● Coordination requires the agent to predict behavior of the other agents

in the system – requires good models
● Coord – Competition – Negotiation
Cooperation – Planning – Centralized Planning
Distributed Planning
● Coherence – achieve global coherence without centralized control
○ Economic markets are good at determining price, but not necessarily provides
optimal resource allocation
 Message Types
● Agents must be capable to participate in a dialog
○ Potential agent role
■ Active, passive or both
■ Master, slave or peer
○ Passive agents
■ Accept info (assertions),
■ Accept query and send a reply (assertion)
■ From the comm network there is no difference between an unsolicited assertion or an
assertion in reply to a query
○ Active agents
■ Issue queries, issue assertions
■ Control subagents, Monitor environment
● Speech act theory is the basis of the inter agent communication
○ Views natural language as actions – requests, suggestions, commitments, and replies.
○ Locution – the spoken (physical) utterances
○ Illocution – intended meaning of the spoken utterance
○ Perlocution – action resulting from the locution
Agent Capabilities

Basic Passive Active Peer

Agent Agent Agent Agent

Receives assertions ∙ ∙ ∙ ∙
Receives queries ∙ ∙
Sends assertions ∙ ∙ ∙
Sends queries ∙ ∙

Table 2.3 from [1]

Inter Agent Message Types
Communicative Action Illocutionary Force Expected Result
Assertion Inform Acceptance
Query Question Reply
Reply Inform Acceptance
Request Request
Explanation Inform Agreement
Command
Permission Inform Agreement
Refusal Inform Agreement
Offer/Bid Inform Agreement
Acceptance
Agreement
Proposal Inform Offer/Bid
Confirmation
Retraction
Denial

Table 2.4 from [1]

Communication Protocol

● Typical levels
○ Lowest level: interconnection
○ Middle level: format, syntax of the info being transferred
○ Top level: meaning or semantics of the info
● Binary: one sender, one receiver
● N-ary: Broadcast, Multicast – one sender, N receivers
● Data structure
○ Sender
○ Receiver (s)
○ Languages in the protocol
○ Encoding and decoding functions
○ Actions to be taken by the receiver
KQML: Inter Agent & Agent Program
Communication

KQML KQML

Agent Application
Agent Program
Knowledge Interchange Format

● Formal syntax for knowledge representation

● Example: Block A on Block B.
(tell
:sender Agent1
:receiver Agent2
:language KIF
:ontology Blocks-World
:content (AND (Block A) (Block B) (On A B)))
Nested KQML
(forward
:from Agent1
:to Agent2
:sender Agent1
:receiver Agent3
:language KQML
:ontology KQML-ontology
:content (tell
:sender Agent1
:receiver Agent2
:language KIF
:ontology Blocks-World
:content (AND (Block A) (Block B) (On A B))))

Forward:from = Content:sender =
Agent1
Forward:to = Content:receiver =
Agent2
Forward:receiver = Agent3
KQML performatives
● Basic query (evaluate, ask-one, ask-all,…)
● Multiresponse query (stream-in, stream-all,..)
● Response (reply, sorry,…)
● Generic info (tell, achieve, cancel, untell, unachieve,
…)
● Generator (standby, ready, next, rest,…)
● Capability-definition (advertise, subscribe, monitor,…)
● Networking (register, unregister, forward, broadcast,…)
Knowledge Interchange Format (KIF)

● Proposed standard for intelligent agents, expert systems, databases…

● Examples
(salary xxx-yy-zzzz designer 40000)
(> (* (width chip1) (length chip1))
(* (width chip2) (length chip2)))
Example KQML dialog

(evaluate
:sender A :receiver B
:language KIF :ontology motors
:reply-with q1 :content (val (torque m1)))

(reply
:sender B :receiver A
:language KIF :ontology motors
:in-reply-to q1 :content (= (torque m1) (scalar 12 kgf)))

● Note: q1 is query reference number

From Figure 8.2 in [2]

 Example KQML dialog
(stream-about
:sender A :receiver B
:language KIF :ontology motors
:reply-with q1 :content (val (torque m1)))
(tell
:sender B :receiver A
:in-reply-to q1 :content (= (torque m1) (scalar 12 kgf)))
(tell
:sender B :receiver A
: in-reply-to q1 :content (= (status m1) normal))
(eos
:sender B :receiver A
: in-reply-to q1)

●stream-about: S wants all relevant answers in R’s Virtual Knowledge Base. Output is
streamed.
●eos ends the stream
From Figure 8.2 in [2]
FIPA ACL

(inform
:sender agent1
:receiver agent2
:content (price good2 150)
:language s1
:ontology hpl-auction
)

● FIPA=Foundation of Intelligent Physical Agents

● Message structure is similar to KQML
● Message attribute fields is also similar to KQML
● FIPA performatives are different than KQML
○ 20 performatives
○ Inform tells the receiver to believe the message content; implies that sender also believes
the message
Ontologies

● Specification of the objects, concepts and relationships

○ In the Block World example, BLOCK represents a concept and ON is a relationship
● Each agent must represent its knowledge using the vocabulary of a specific
ontology
● Note KQML – specifies the Ontology
● How are Ontologies co-ordinated?
Ontologies for ACL

● Represent part of the world.

● Shared virtual world, which provide the terms for communication.
● If 2 agents agree on the upper nodes of a taxonomy, then these agents can
focus on the language content in this context.
Agent Interaction Protocols

● Interaction implies conversation

● Objective:
○ Ensure achievement of overall goals,
○ Coherence across agents
○ Retain agent autonomy
● Shared goals, common tasks, reduce conflicts, pool knowledge and evidence
Agent Interaction Protocols (contd)

● Co-ordination
● Co-operation
● Contract Net
● Blackboard
● Negotiation
● AIGA
Co-ordination Protocols

● Reason for co-ordination: better utilization of resources,

avoid duplication, maintain coherence
○ Timely updates, agent synchronization
○ Distribute control and data
■ Disadv: System state is distributed
■ Adv: Reduce points of bottleneck
○ Goal graph
■ Relates goals and resources
■ Identifies dependencies
■ Assignment of goals to agents
■ Manage graph traversal and report results
Co-operation Protocol

● Supports decomposition and distribution of tasks (divided and

conquer)
○ Avoid overloading of critical resources
○ Task assigned to agents with matching skills
○ Master / slave relationships in task assignment
○ Minimize communication and synchronization cost: spatial and
semantic proximity
○ Migrate tasks if necessary
○ Redundancy and fault tolerance
● Task distribution approaches
○ Market mechanisms: price, utility
○ Contract net
○ Credibility, belief management
○ Static plans: task to resource map
 Contract Net

● Manager wants to find contractors

○ Announce task
○ Receive bids
○ Award contracts
○ Results,
task accomplishment
● Contractors' role
○ Evaluate capability
○ Respond (Bid, No Bid)
○ Perform task
○ Report results
● Manager assess contractor capability
● Distributed computing implications
Blackboard

● Specialized Knowledge Sources (KS)

● Multiple approaches to problem solution
● Knowledge representation is KS controlled
● Control transfer
● Data exchange
● Synchronization
Blackboard Architecture

Executin Library
Blackboar g Of
d Activated KSs
KS

Event
s
Pending
Control KS
Componen Activation
ts s

Distributed Memory Systems?

Granuality?
Negotiation

● Goes through a cycle

○ Offer
○ Evaluation
○ Identify agreements
○ Identify disagreements
○ Repeat with counter offer
● Granularity, efficiency, stability