Ask Kalicube

Indexing Annotation Extraction Quality forms Level 5 of Jason Barnard's Indexing Annotation Hierarchy. These five annotations determine HOW a selected chunk is deployed in AI-generated outputs: (1) Sufficiency - does the chunk fully answer the likely question? (2) Dependency - does understanding require adjacent chunks or external context? (3) Standalone Score - can the chunk be extracted and used independently? (4) Entity Salience - is the focus entity CENTRAL to the chunk or mentioned in passing? (5) Entity Role - is the entity the subject, object, example, or authority source? Unlike multipliers that rank, Extraction Quality annotations govern usage: a highly-ranked chunk with poor extraction quality might be summarized rather than quoted, contextualized rather than standalone, or used as supporting evidence rather than primary source.

Extraction Quality is the deployment layer. You've survived the gates, been defined, routed correctly, and ranked highly. Now: HOW does your content appear in the AI's response? Are you quoted directly or paraphrased? Used standalone or needing context? Cited as the authority or mentioned as supporting evidence? High Extraction Quality means your exact words reach the user. Low Extraction Quality means the AI rewrites you - and you lose control of the message.

The rise of AI-generated responses changed what "visibility" means. Ranking #1 used to mean users saw YOUR content. Now, ranking #1 might mean AI summarizes your content into ITS response. Extraction Quality annotations determine this distinction. High sufficiency and standalone scores mean quoted verbatim. High entity salience as subject means authoritative source. Understanding Extraction Quality is understanding how to remain visible even when AI mediates the message.