Independence assumptions considered harmful
Alexander M. Franz · 1997
Many current approaches to statistical language modeling rely on independence assumptions between the different explanatory variables. This results in models which are computationally simple, but which only model the main effects of the explanatory variables on the response variable. This paper presents an argument in favor of a statistical approach that also models the interactions between the explanatory variables. The argument rests on empirical evidence from two series of experimetns concerning automatic ambiguity resolution.