"English"-English phrase table

"English"-English phrase tables are constructed as follows.

1. Parallel Corpus

   We prepare a French-English parallel corpus.

2. Pattern-Based Machine Translation

   We use a French-English PBMT. Thus, we obtain "English" sentences from French sentences. These "English" sentences are pairs of English sentences.

3. "English"-English phrase tables

   We construct "English"-English phrase tables using Giza++ [6] and training-model.perl [4] from "English" sentences (outputs of a French-English PBMT) and English sentences (from parallel corpus).

Figure 3 shows a flow chart for constructing "English"-English phrase tables.

図: Flowchart for constructing "English"-English Phrase Tables