01 / Definition Term

= Summarization — compression of information while preserving meaning

= NOT just shortening — selecting, classifying, and restructuring

02 / Three Types Taxonomy

Extractive

= select key sentences from source

+ preserves original wording

- fragments context, misses implicit relationships

. tool: traditional NLP, highlighters

Abstractive

= rewrite in new words, shorter

+ can capture gist better than original

- risk of hallucination (introducing false claims)

. tool: LLMs, human paraphrasing

Analytical (CLN)

= decompose into typed semantic lines

+ preserves meaning AND structure

+ scannable: operator column reveals information types

+ forces author to classify every fact

+ reusable: grep by operator, cross-reference by topic

- requires knowledge of operator set

. tool: CLN notation

03 / Why Analytical Wins Cognition

. reading prose summary → passive consumption → forget

. creating CLN summary → active classification → encode

* the cognitive cost of structuring IS the learning benefit

04 / AI & CLN Practice

! AI-generated summaries without structure feed ai-amnesia

+ AI-generated CLN summaries + human refinement = optimal

05 / See Also

@ see: cln, CLN-memorization, info-chunking

@ template: ^summarize (CLN spec v1.2)

---