To accurately recognize data even when handwriten character data to be inputted lacks strokes that originally existed, to suppress a recognition dictionary from becoming fat, and also to reduce recognition time.

Line segmenting processing that divides inputted ink data into plural line segments is performed (step 102), the line segments are partially eliminated from plural line segments according to a prescribed line segment elimination rule (step 103), line segments are added to ink data whose partial line segments are eliminated based on a prescribed line segment addition rule (step 104), a recognition dictionary is retrieved based on ink data to which the line segments are added and a candidate character is extracted (step 106).