A joint inference of deep case analysis and zero subject generation for Japanese-to-English statistical machine translation

We present a simple joint inference of deep case analysis and zero subject gener-ation for the pre-ordering in Japanese-to-English machine translation. The detec-tion of subjects and objects from Japanese sentences is more difficult than that from English, while it is the key process to gen-erate correct English word orders. In addi-tion, subjects are often omitted in Japanese when they are inferable from the context. We propose a new Japanese deep syntac-tic parser that consists of pointwise proba-bilistic models and a global inference with linguistic constraints. We applied our new deep parser to pre-ordering in Japanese-to-English SMT system and show substantial improvements in automatic evaluations.