Text Summarization: From Extractive Methods to Abstractive LLM-Based Summarization

Status: public · Confidence: medium (0.78) · Basis: verified_sources

## TL;DR
Text summarization condenses documents into shorter outputs. The evidence-backed story runs from pointer-generator models to pretrained sequence-to-sequence systems such as BART and PEGASUS.

## Core Explanation
Extractive summarization selects existing sentences or spans. Abstractive summarization may generate new wording, which can make summaries more fluent but also creates faithfulness risk.

## Detailed Analysis
For a public knowledge base, the safest claims describe documented model mechanisms and avoid broad claims that summaries are accurate by default. Faithfulness, attribution, and domain reliability remain the central quality issues for modern summarization systems.

## Related Articles

- [Text Classification: Zero-Shot, Few-Shot, and LLM-Based Document Categorization](../text-classification.md)
- [AI for Mental Health: LLM-Based Therapy, Digital Interventions, and Clinical Trials](../ai-for-mental-health.md)
- [AI Language Translation and Interpretation: LLM-Based Translation, Simultaneous Interpretation, and Quality Estimation](../ai-language-translation-interpretation.md)