A visual guide to understanding retrieval architectures, data fit, and implementation tradeoffs.
Retrieves information based on conceptual similarity (Vectors/Distance).
Text is broken down into raw, flexible chunks.
Retrieves information based on explicit relationships and hierarchies.
Data must be structured into strict nodes and edges.
Nuance is lost in extraction.
Angry tones and unstructured emotion.
Requires natural flow, not strict relationships.
Naturally structured as clear networks.
Strict rules, entities, and definite links.
Functions, dependencies, and rigid syntax.
Different methodologies for traversing the knowledge graph.
Answers specific questions.
Answers broad questions by analyzing clusters.
Combines both methodologies.