GLOSSARY

RAG & enterprise AI glossary

Plain-language definitions of the terms behind secure, permission-aware retrieval.

A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
A

Audit Log

A tamper-evident record of queries and responses, used for traceability, governance and review.

C

Chunking

Splitting documents into smaller passages so they can be embedded and retrieved precisely.

Citation

A reference attached to a generated answer pointing back to the source passage it came from.

Compliance Gate

A checkpoint that documents must pass before they are indexed, enforcing governance rules.

Confidence Score

An indication of how well-supported an answer is by the retrieved source material.

Connector

An integration that lets Vijnara read content from a system such as SharePoint, Drive or Slack.

D

Data Residency

The requirement that data be stored and processed within a specific geographic region.

E

Embedding

A numeric vector representation of text that captures meaning for semantic search.

G

Grounding

Constraining a model's answer to retrieved, permitted context rather than its general training data.

H

Hallucination

When a model produces confident but unsupported or fabricated information.

Hybrid Retrieval

Combining semantic (vector) and keyword search to improve precision and recall.

L

LLM

Large Language Model — the model that composes natural-language answers from retrieved context.

M

Maker-Checker

A workflow where one person submits content and another reviews and approves it before it goes live.

Multi-Tenancy

Architecture that keeps each organization's data isolated within a shared platform.

P

Permission-Aware Retrieval

Retrieval that filters results by each user's role and department access before content is read.

R

RAG

Retrieval-Augmented Generation — retrieving relevant content and using it to ground a generated answer.

Reranking

Reordering retrieved passages by relevance before they are passed to the model.

Role-Based Access Control

Granting permissions based on a user's role, so access follows responsibility.

S

Semantic Cache

A cache of prior answers keyed by meaning, used to speed up repeated or similar queries.

Semantic Search

Search that matches on meaning rather than exact keywords, powered by embeddings.

Sub-Processor

A third party engaged to process data on the platform's behalf under contractual safeguards.

V

Vector Store

A database that indexes embeddings for fast similarity search across documents.