The intellectual granularity that gets from inside the corpus is smaller, more exquisite to the depict of natural language, true and accurate.

As a result of large-scale the price that gets language knowledge is very big, wraparound to accidence, grammar and semantic regular collection hard comprehensive, the function of machine translation system hesitates to press forward all the time. Use the bilingual corpus resource that has existed to provide experience for new interpreter requirement, it is one of important ways that improve quality of translation of machine translation system at present. EBMT has very distinct effect to the interpreter of identical or similar text version, as the addition of the dimensions of illustrative sentence library, its effect is more and more distinct also. To example library in already some text, can achieve the interpreter result of high quality directly. To very as similar as the example that exists in example library text version, can carry analogy deduction, undertake to translating a result a few revise, construction translates a result approximately. The interpreter that is based on example has numerous advantage, differ in thousands ways is on specific implementation, a lot of places still have quite large latent capacity, because this is one of heat of the research of machine translation all the time in recent years. But the limitation as a result of corpus dimensions, the machine translation that is based on example is very inaccessible very tall match rate, consequently, so far still rarely machine translation system uses the dinkum method that is based on example, it is one of regarding much interpreter engine as the machine translation that is based on example commonly, in order to raise correct rate of the interpreter.