A Simplified Heuristic Version of Raviv's Algorithm for Using Context in Text Recognition

Word-posi t ionindependent and word -pos i t i on dependent n-gram p r o b a b i l i t i e s were est imated from a la rge Engl ish language corpus. A t e x t r e c o g n i t i o n problem was s imu la ted , and using the est imated n-gram p r o b a b i l i t i e s , four experiments were conducted by the f o l l o w i n g methods of c l a s s i f i c a t i o n : w i thou t contex tua l i n f o r m a t i o n , Rav iv 's recu rs i ve Bayes a l go r i t hm , the modi f ied V i t e r b i a l g o r i t h m , and a proposed h e u r i s t i c approximation to Rav iv ' s a l go r i t hm . Based on the est imates of the p r o b a b i l i t i e s o f m i s c l a s s i f i x a t i o n observed in the four exper iments, the above methods are compared. The h e u r i s t i c approximat ion of Rav iv 's a lgo r i t hm performed j u s t as w e l l as Rav iv 's and requ i red f a r less computat ion.