This system is an experiment of examples based approach. It is based on a corpus containing 220 thousand sentence pairs with word alignment. The system contains four parts: matching and search, fragment matching, fragment assembling, evaluation and post processing. We use word alignment information to find and combine fragments.