Skip to content

Glossary

A

Anthropic
LLM provider offering Claude models. Supported as an alternative to WatsonX.

C

Chain of Thought (CoT)
Advanced reasoning technique that breaks complex questions into sub-questions, answers each step, and synthesizes comprehensive answers with source attribution.
ChromaDB
Supported vector database option for storing document embeddings.
Collection
Logical grouping of documents for organization and querying.
ConversationRepository
Unified repository pattern for managing conversation sessions, messages, and summaries.

D

Docling (IBM)
Advanced document processing library for extracting content from PDFs, images, tables, and complex formats.

E

Elasticsearch
Supported vector database option with hybrid search capabilities.
Embedding
Vector representation of text that captures semantic meaning for similarity search.

G

GHCR (GitHub Container Registry)
Container registry where RAG Modulo Docker images are published.

L

LLM (Large Language Model)
AI model used for generating responses based on retrieved context.

M

Material Theme
Modern documentation theme used for GitHub Pages with dark/light mode support.
Milvus
Default vector database for RAG Modulo, optimized for similarity search at scale.
MkDocs
Static site generator used for project documentation.

O

OpenAI
LLM provider offering GPT models. Supported as an alternative to WatsonX.

P

Pinecone
Cloud-based vector database option for managed vector search.
Pipeline
Automated workflow for document processing and query execution.

R

RAG (Retrieval-Augmented Generation)
AI technique combining document retrieval with LLM generation for accurate, grounded answers.
Reranking
Post-processing step that improves search relevance by reordering retrieved documents.

S

Source Attribution
Feature that tracks and displays which documents/chunks were used to generate answers.

T

Token
Unit of text processed by LLMs. RAG Modulo tracks token usage for cost monitoring.
TruffleHog
Secret scanning tool used in CI/CD to detect exposed API keys.

V

Vector Database
Specialized database for storing and searching high-dimensional embeddings.

W

WatsonX
IBM's LLM platform, the default provider for RAG Modulo.
Weaviate
Supported vector database with semantic search capabilities.

For more detailed explanations, see the relevant sections in the documentation.