Neural machine Translation
Statistical Machine Translation
Syntax-based models are trained on automatically extracted parallel corpora, with automatically generated word alignments and linguistic annotations. To overcome all these sources of noise, there have been efforts to manually create parallel treebanks.
Parallel Treebanks is the main subject of 9 publications. 4 are discussed here.
Topics in SyntaxBasedModelsInversion Transduction Grammars | Hierarchical Models | String To Tree | Tree To String | Tree To Tree | Binarizing SCFG | Generative Syntax Models | Dependency Structure | Deep Syntax | Parse Quality | Parallel Treebanks | Syntactic Coherence