Overview
Direct Answer
Extractive summarisation is a Natural Language Processing technique that automatically condenses documents by selecting and retaining the most salient sentences from the original source, preserving their exact wording without paraphrase or generation of new content.
How It Works
The approach ranks sentences using statistical or machine learning methods—such as term frequency-inverse document frequency (TF-IDF), graph-based algorithms, or neural scoring models—to identify those carrying the greatest semantic importance. Selected sentences are then assembled in their original sequence to form a shorter document, maintaining coherence through preservation of the source text's structure and language.
Why It Matters
Organisations benefit from rapid document processing at scale, particularly where speed and interpretability are critical; since no novel text is generated, output remains fully traceable to source material, supporting compliance, auditability, and stakeholder trust. This approach reduces computational overhead compared to abstractive methods, making it cost-effective for high-volume document workflows.
Common Applications
Applications include legal document review, where key clauses and obligations must be flagged; news aggregation platforms requiring fast headline extraction; customer support ticket prioritisation; and scientific literature filtering in research institutions seeking rapid assessment of publication relevance.
Key Considerations
The technique cannot bridge gaps in source content or reshape information for clarity, limiting its effectiveness where documents are poorly structured or where context requires paraphrasing. Quality depends heavily on sentence-ranking algorithm selection and may miss nuanced information valuable to specific user contexts.
More in Natural Language Processing
Seq2Seq Model
Core NLPA neural network architecture that maps an input sequence to an output sequence, used in translation and summarisation.
Tokenisation
Semantics & RepresentationThe process of breaking text into smaller units (tokens) such as words, subwords, or characters for processing by language models.
Multilingual Model
Semantics & RepresentationA language model trained on text from dozens or hundreds of languages simultaneously, enabling cross-lingual understanding and generation without language-specific fine-tuning.
Semantic Search
Core NLPSearch technology that understands the meaning and intent behind queries rather than just matching keywords.
Abstractive Summarisation
Text AnalysisA text summarisation approach that generates novel sentences to capture the essential meaning of a document, rather than simply extracting and rearranging existing sentences.
Sentiment Analysis
Text AnalysisThe computational study of people's opinions, emotions, and attitudes expressed in text.
Grounding
Semantics & RepresentationConnecting language model outputs to real-world knowledge, facts, or data sources to improve factual accuracy.
Slot Filling
Core NLPThe task of extracting specific parameter values from user utterances to fulfil a detected intent, such as identifying dates, locations, and names in booking requests.