CourseNatural Language Processing: The Complete Course · Module 8: Extraction, Question Answering and Summarisation · part 56 of 73
Part 56 · Module 8: Extraction, Question Answering and Summarisation

Lesson 3: Summarising long email threads for Rosa

47 min read·14 Oct 2026
Module 8 · Lesson 3 · About 55 minutes · Needs: basic Python; Module 6 Lesson 3 (sentence embeddings) helps; Module 7 (transformers) helps but is not required

The situation

It is Wednesday 28 October 2026, project week 8. Rosa does not read long threads; she reads the first lines and the last. She has asked for a summary of every long thread on her Monday list, three sentences at most, and she wants to trust every fact in it.

The running example is TH01, Mrs Joan Hartley's thread about her Volt 2 battery (simulated): 11 emails from 10 June to 19 July 2026, three chases, each angrier than the last, and an apology from Rosa. Ines wrote a reference summary by hand.

TH01: 11 EMAILS, 10 JUNE TO 19 JULY 2026 10 Jun 11 12 15 6 Jul 8 13 14 17 18 19 three chases Rosa Rosa wants: three sentences, every fact true
The Hartley thread runs for 39 days: customer emails (white), staff replies (grey), three chases that grow angrier (red) and Rosa's apology (green); Rosa wants it in three true sentences.

The rest of this part opens with an account

Signing up is free and takes one email. Nothing to pay, and nothing renews.