RBMT as an alternative to SMT for under-resourced languages
Guillaume de Malézieux, Amélie Bosc, Vincent Berment · 2014
Despite SMT (Statistical Machine Translation) recently revolutionised MT for major language pairs, when addressing under-resourced and, to some extent, mildly-resourced languages, it still faces some difficulties such as the need of important quantities of parallel texts, the limited guaranty of the quality, etc. We thus speculate that RBMT (Rule Based Machine Translation) can fill the gap for these languages. 1