The COMPEL Glossary Graph visualizes relationships between framework terminology, showing how concepts interconnect across domains, stages, and pillars. Term nodes cluster by pillar affiliation while cross-references reveal semantic dependencies — for example, how risk appetite connects to control effectiveness, model governance, and assurance requirements. This network representation helps practitioners navigate the framework vocabulary and understand that COMPEL terminology forms a coherent conceptual system rather than isolated definitions.
COMPEL Glossary / late-chunking
Late chunking
A chunking pattern in which the full document is embedded as a long sequence and chunk boundaries are applied at retrieval time against the embedded representation — rather than chunking before embedding.
What this means in practice
Preserves long-range context in the embedding while still returning chunk-sized results to the generator.
Synonyms
retrieval-time chunking , late-interaction chunking
See also
- Chunking — The process of dividing documents into units — typically fixed-token windows or paragraph-level segments — suitable for embedding and retrieval.
- Semantic chunking — A chunking strategy that respects semantic boundaries — sentence, paragraph, or topic-shift — rather than fixed token windows.
- Embedding model — A model that maps text, images, or multimodal content to dense vector representations used for retrieval, clustering, and similarity search.