A parametric adjustment method for statistical machine translation

A parametric adjustment method for statistical machine translation, to a parameter adjustment method. To solve the problem OVERFITTING existing methods likely to cause parameter in the parameter adjustment process and falling into local optimal solution. Method: Step 1: using monolingual corpora desired translation language model construct; Step 2: Using parallel corpus constructed translation model; Step three: the objective function to adjust the parameters. The method of the present invention is easy to implement, and can prevent over-fitting in the parameter adjustment process, the number and combination of features more. Further, since the objective function is convex, so the training process also possible to obtain the global optimal solution. The method of the present invention to adjust the translation parameters for the system.