Loading timeline…
19901990s
Statistical Machine Translation
Translating text based on statistical models derived from bilingual text corpora.
Why It Was Important
Pioneered by researchers at IBM (the Candide project), SMT replaced linguist-written grammatical rules with statistical probabilities. If the engine saw 'chien' translated to 'dog' 99% of the time in millions of parsed documents, it learned the translation. This data-driven approach eventually powered Google Translate in its early years.
Who Invented It
IBM Thomas J. Watson Research Center
A team led by Peter Brown prioritizing data over linguistic theory.
Applications
- Online Translation Services (early)
- Corpus Linguistics
- Document Parsing
Key Papers
- A Statistical Approach to Machine Translation
Peter F. Brown et al. · Computational Linguistics · 1990