next up previous contents index
Next: Absolute discounting and backing-off Up: Language model smoothing: modelling Previous: Linear discounting and backing-off

Linear interpolation


In linear interpolation, a weighted average between the relative frequencies tex2html_wrap_inline45737 and the general distribution tex2html_wrap_inline45703 is computed:


In other words, the difference to linear discounting  in connection with backing-off  is that the more general distribution tex2html_wrap_inline45703 is used in all cases [Jelinek & Mercer (1980), Nadas (1984)]. The mathematical framework becomes rather complex for estimating the unknown parameters in linear interpolation. In most cases, the so-called EM algorithm  is used as described in the appendix; EM stands for expectation-maximisation.      

EAGLES SWLG SoftEdition, May 1997. Get the book...