Package com.tnsai.models.agent

package com.tnsai.models.agent

BDI (Belief-Desire-Intention) model classes for agent reasoning.

This package implements the BDI architecture, a well-established model for rational agents based on folk psychology. BDI agents reason about the world using three key mental attitudes:

BDI Mental States

 ┌─────────────────────────────────────────────────────────────┐
 │                    BDI Reasoning Cycle                       │
 ├─────────────────────────────────────────────────────────────┤
 │                                                              │
 │     BELIEFS ──────────► DESIRES ──────────► INTENTIONS      │
 │    (Knowledge)         (Goals)            (Commitments)      │
 │         │                  │                    │            │
 │         │                  │                    ▼            │
 │         │                  │               ┌────────┐        │
 │         │                  └──────────────►│  PLAN  │        │
 │         │                                  └────┬───┘        │
 │         │                                       │            │
 │         │                                       ▼            │
 │         │                                  ┌────────┐        │
 │         └─────────────────────────────────►│ ACTION │        │
 │                                            └────────┘        │
 └─────────────────────────────────────────────────────────────┘
 

Model Classes

Class	Role	Example
Belief	Knowledge about the world	"user_language": "Turkish"
Desire	Goals to achieve	"Provide accurate information"
Intention	Committed desires	"Research the topic thoroughly"
Plan	Strategy to fulfill intentions	"1. Search, 2. Analyze, 3. Summarize"
Capability	Skills the agent has	"Web search", "Data analysis"
Communication	Communication style	Formal, Verbose
AgentIdentity	Agent metadata	name, description, version

Usage Example

 public class ResearchAgent extends Agent {
     @Override
     protected List<Belief> getBeliefs() {
         return List.of(
             new Belief("domain", "scientific research"),
             new Belief("sources", "academic databases, journals")
         );
     }

     @Override
     protected List<Desire> getDesires() {
         return List.of(
             new Desire("Provide well-researched answers", 10),
             new Desire("Cite sources accurately", 9)
         );
     }

     @Override
     protected List<Intention> getIntentions() {
         return List.of(
             new Intention("Always verify facts before responding")
         );
     }

     @Override
     protected List<Plan> getPlans() {
         return List.of(
             new Plan("Deep Research",
                 "Comprehensive investigation",
                 "1. Search multiple sources\n2. Cross-reference\n3. Synthesize")
         );
     }
 }
 

LLM Integration

BDI models contribute to the system prompt, providing context that guides the LLM's reasoning and responses.

See Also:

• Belief
• Desire
• Intention
• Plan
• Capability
• Agent

Related Packages

Package	Description
com.tnsai.models	Data models for the TnsAI framework.
com.tnsai.models.chat	Multimodal content parts for LLM communication.
com.tnsai.models.role	Gaia methodology model classes for role-based agent design.

Class	Description
AgentIdentity	Represents a single identity trait or personality characteristic of an agent.
Belief	Represents a single belief that an agent holds about the world.
BeliefRevisionStrategy	Strategy interface for handling conflicting beliefs.
Capability	Represents a capability or skill that an agent possesses within the BDI architecture.
Communication	Represents the communication style and persona of an agent.
Desire	Represents a desire or goal that an agent wants to achieve.
DesireConflictResolver	Strategy interface for resolving conflicts between desires with the same priority.
ErrorRecoveryScenario	Represents an error recovery scenario with strategy.
Intention	Represents an intention - a committed task that the agent is actively working on.
Plan	Represents a strategic plan or approach within the BDI architecture.