C H A P T E R 9
About the Recognition System
multiple stroke units into meaningful groups. For example, certain letters (such as
an uppercase E) might be composed of multiple strokes. The process of grouping
input strokes is influenced by the user preference settings for handwriting style and
The recognizer that claimed one or more stroke units returns to the view one or
more interpretations of the strokes. The gesture and shape recognizers return
only one interpretation to the view. The text recognizer usually returns multiple
interpretations to the view.
Associated with each interpretation is a value, called the score, which indicates
how well the input matched the system-defined model used by the recognizer that
interpreted it. When multiple recognizers are enabled, the system selects the best
interpretations based on their scores and the application of appropriate heuristics.
For example, the text recognizer might choose between interpreting a stroke as a
zero or as the letter O based on whether you have specified that the view accepts
numeric or alphabetic input.
The recognizer that claimed the strokes places its best interpretations in another
kind of unit that is returned to the view. The text recognizer returns word units, the
shape recognizer returns shape units, and the gesture recognizer returns gesture
. Each of these units contains data representing one or more strokes. A word
unit represents a single recognized word, a shape unit represents a single
recognized shape, and a gesture unit represents a single recognized gesture, as
shown in Figure 9-1. The next several sections describe how the system handles
each of these units.
When the recognition system returns a gesture unit to the view, the view performs
the action associated with that gesture automatically. The action taken is dependent
on the kind of view that received the gesture unit.
Edit views and paragraph views respond automatically to system-defined gestures
such as scrubbing items on the screen, adding spaces to words, selecting items on
the screen, and so on. Other kinds of views may do nothing in response to a
You can provide an optional
method to take action in
response to any standard gesture. For example, you can use this method to respond
to gestures in views that are not paragraph views or edit views. You can also use
this method to override or augment the standard behavior of a particular view in
response to system-defined gestures. At present, you cannot define custom gestures
to the system.