Revision as of 21:27, 30 January 2020 edit Citation bot (talk \| contribs) Bots 5,866,681 edits m Alter: journal, issue. Add: bibcode, volume, arxiv, doi, date. \| You can use this bot yourself. Report bugs here. \| Activated by User:Sebaskohl \| Category:Machine_translation \| via #UCB_Category ← Previous edit		Revision as of 22:14, 12 April 2020 edit undo ChrisBrew (talk \| contribs) 5 edits changed the sentence including 'for almost twenty years' to better reflect situation in 2020, by which time practical MT is mostly neural, and doesn't rely on IBM alignments to the same extent. Next edit →
Line 1: '''IBM alignment models''' are a sequence of increasingly complex models used in [[statistical machine translation]] to train a translation model and an alignment model, starting with lexical translation probabilities and moving to reordering and word duplication.<ref>{{cite web \| url = http://www.statmt.org/survey/Topic/IBMModels \| title = IBM Models \| date = 11 September 2015 \| publisher = SMT Research Survey Wiki \| access-date = 26 October 2015}}</ref> They ~~have~~ underpinned the majority of statistical machine translation systems for almost twenty years starting in the early 1990s, until [[neural machine translation]] began to dominate. These models offer principled probabilistic formulation and (mostly) tractable inference.<ref>{{cite web \| url = http://mlg.eng.cam.ac.uk/yarin/PDFs/PY-IBM_presentation.pdf \| authors = Yarin Gal, Phil Blunsom \| title = A Systematic Bayesian Treatment of the IBM Alignment Models \| date = 12 June 2013 \| publisher = University of Cambridge \| access-date = 26 October 2015}}</ref> The original work on statistical machine translation at [[IBM]] proposed five models, and a model 6 was proposed later. The sequence of the six models can be summarized as:

IBM alignment models: Difference between revisions