<p>An efficient and simple approach to encoding ideographic characters as sequences of input strokes or stroke categories is shown in Figure 5. Each character (53) is represented by one or more sequences of one or more components (55). Each component corresponds to a plurality of alternative stroke sequences, each of which is associated with a probability that it will be the sequence which the user enters to specify the given component or character. The probability associated with the user's preferred stroke sequence is automatically increased by the system when the character is selected, thus, automatically adapting to a user's preferences.</p>