C. Integration of Rules and Statistic
The result from statistical method is sentence-break or nonsentence-break for every space. As described in section II,
Thai Grammar Rule can identify some spaces as nonsentence-break, so this can minimize a number of wrong
sentence-break from statistical prediction. At run time, rules
are applied to each statistical result before predicting the next
space in order that some features – NWrd_PrevSB and V –
which depend on previous results are to be calculated
correctly