Building Translation Memory System by N-gram

摘要

In this paper, we proposed a method that built translation memory system by N-gram. It doesn't use any lingual analytical technologies such as chunking, target-text generator, word alignment and so on which are often used in other similar translation memory systems; besides, it has a high proposal speed. These virtues make it is easy to build a translation memory system for those languages that lingual analyses are difficult to progress. We evaluate our system from aspect of assistant efficiency, and experiments show system got a high speed performance; At last, we give some theory compares between our translation memory system and other similar systems.