This week
- Bench
- 17 March 2027 Find the seam locating the join
- Slides
- Open the deck
A rule-based system breaks in a way that looks broken. A statistical one breaks in a way that looks like English for six words and then turns.
The reason is where the text came from. Parliamentary proceedings and legislative records were among the few large, sentence-aligned bilingual corpora available, so the systems learned the register of committee minutes and applied it to menus.
Outline#
- parallel corpora: what was available, and what that made the systems good at
- the phrase table as a suspect with a memory but no understanding
- locating a join, and what its position tells you about unit length
- dating an artefact from the shape of its errors
Teaching team
- Aurel Marchetti — tutor