Skip to content

AICO Terminology Glossary

Core Terms

Communication

  • User Input: Raw input (text, voice, image, multimodal)
  • User Message: Processed, structured input ready for system processing
  • Request: API-level HTTP/WebSocket request from frontend to backend
  • Query: Specific information request within user input
  • Conversation: Complete ongoing relationship/communication context
  • Session: Technical connection/authentication period
  • Thread: Logical grouping of related exchanges
  • Turn: One side of communication (user OR AICO)
  • Exchange: Complete back-and-forth (user turn + AICO turn)
  • Response: AICO's reply (text, voice, multimodal)
  • Reaction: AICO's behavioral/emotional response

System Components

  • Conversation Engine: Core orchestration for conversation flow, context integration, multimodal processing
  • Conversation Manager: Handles conversation state, thread management, context switching
  • Response Generator: LLM-based component producing textual responses
  • Input Processor: Handles and routes different input types

Context & Memory

  • Context: Current situational information influencing conversation
  • Conversation Context: Historical and current conversation state
  • User Context: User's situation, mood, preferences
  • Thread Context: Specific context within conversation thread
  • Memory: Persistent conversation and user information

Message Bus Topics

  • Input: user/input/{type} (text, voice, multimodal)
  • Conversation: conversation/{action} (start, message, response, end)
  • Context: context/{type} (user, emotional, social, personality)
  • Response: response/{stage} (generated, enhanced, delivered)

Usage Guidelines

When to Use Each Term

Conversation vs Chat: - Use "conversation" for the broader, ongoing relationship context - Use "chat" only when referring to legacy UI components or casual text-based interaction - Component name: "Conversation Engine" (reflects full orchestration capabilities)

Input vs Message vs Request: - "User Input": Raw data from user interface - "User Message": Structured, processed input ready for system handling
- "Request": Technical API/HTTP layer communication - "Query": When user is asking for specific information

Session vs Thread vs Turn vs Exchange: - "Session": Technical authentication/connection period - "Thread": Logical conversation grouping (like email threads) - "Turn": One person speaking (user says something OR AICO responds) - "Exchange": Complete conversation unit (user turn + AICO turn = full back-and-forth)

Response vs Reaction: - "Response": Primary term for AICO's reply to user - "Reaction": Behavioral/emotional response separate from text

Message Bus Topic Patterns

Standard Format: {domain}/{action}/{type}

Examples: 

user/input/text
user/input/voice  
user/input/multimodal
conversation/message/received
conversation/response/generated
conversation/context/updated
context/emotional/current
context/personality/expression
response/enhanced/ready

Consistency Rules

  1. Component Names: Use "Conversation Engine" (reflects full orchestration capabilities)
  2. API Endpoints: Use "conversation" for user-facing endpoints (/api/v1/conversation/)
  3. Message Topics: Follow domain/action/type pattern consistently
  4. Documentation: Use full terms on first reference, abbreviations sparingly
  5. Code: Prefer explicit terms over abbreviations (conversation_context vs conv_ctx)

Terms to Avoid

  • Deprecated: "chat" when referring to the overall conversation system (use "conversation" instead)
  • Ambiguous: "interaction" without qualifier (specify type)
  • Technical: "request" when referring to user input (use "user input" or "user message")
  • Inconsistent: Mixed topic patterns (stick to domain/action/type)

Cross-Reference

  • Memory System: Stores conversation history and context
  • Emotion Simulation: Provides emotional context for conversations
  • Personality Simulation: Influences conversation style and responses
  • Avatar System: Provides visual embodiment during conversations
  • Voice System: Handles speech-to-text and text-to-speech for conversations

External Integrations

  • LLM Integration: Powers response generation in Conversation Engine
  • Multimodal Processing: Handles image/video input in conversations
  • Plugin System: Extends conversation capabilities