GIZA++是德国Och提供的针对统计机器翻译模型的训练工具集。这个工具可以生成IBM统计翻译模型的第4和第5种模型，采用两种方向的训练（比如对同一个语料，从中文到英文和从英文到中文的训练），再配以适当的编码，可以提取出适合基于Phrase的统计机器翻译phrase table来，为后续的工作提供数据支持。-GIZA++ Is provided by Germany Och statistical machine translation model training tool set. This tool can generate IBM statistical translation model of paragraphs 4 and 5 kinds of models, uses two types of orientation training (for example, with a corpus, from Chinese to English and from English to Chinese language training), together with the appropriate codes, can be extracted for Phrase-based statistical machine translation of the phrase table to work for the follow-up data.