Alphanumeric data entry apparatus and method using multicharacter keys of a keypad
First Claim
Patent Images
1. A method for predicting characters input into a reduced-key keypad comprising a plurality of multi-character keys, the method comprising:
- using a processor cooperating with the reduced-key keypad to populate a multi-level rule tree comprising a plurality of rules, each rule predicting a character in a current input word, rules in a respective level predicting a numbered character position in the current input word;
said multi-level rule tree comprising a plurality of branches connecting the plurality of rules together;
each branch associated with a respective two-character sequence in the current input word, an upper level rule for predicting a first character in the respective two-character sequence, and a lower level rule for predicting a second character in the respective two-character sequence, each rule having a counter value representing a number of correct predictions;
for each previously correct prediction of the current input word using a respective rule, using the processor to increment the corresponding counter value in the multi-level rule tree to weight the respective rule;
using the processor to delete from the multi-level rule tree at least one rule having a weight value less than a threshold value; and
using the processor to weight subsequent predictions of other input characters based upon the counter values in the multi-level rule tree.
3 Assignments
0 Petitions
Accused Products
Abstract
A method and apparatus for predicting the desired alphanumeric character of a depressed multi-character key of a reduced-key keypad based upon the previous selection of characters. Rule trees defining the prediction associated with a depressed key based upon the previous selection of characters are dynamically built upon use to readily function with all language styles and vocabularies. Moreover, the rule trees are weighted to increase the probability of predicting the desired character and to be quickly adaptable to different users having different language styles or vocabularies.
42 Citations
22 Claims
-
1. A method for predicting characters input into a reduced-key keypad comprising a plurality of multi-character keys, the method comprising:
-
using a processor cooperating with the reduced-key keypad to populate a multi-level rule tree comprising a plurality of rules, each rule predicting a character in a current input word, rules in a respective level predicting a numbered character position in the current input word; said multi-level rule tree comprising a plurality of branches connecting the plurality of rules together; each branch associated with a respective two-character sequence in the current input word, an upper level rule for predicting a first character in the respective two-character sequence, and a lower level rule for predicting a second character in the respective two-character sequence, each rule having a counter value representing a number of correct predictions; for each previously correct prediction of the current input word using a respective rule, using the processor to increment the corresponding counter value in the multi-level rule tree to weight the respective rule; using the processor to delete from the multi-level rule tree at least one rule having a weight value less than a threshold value; and using the processor to weight subsequent predictions of other input characters based upon the counter values in the multi-level rule tree. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11)
-
-
12. An electronic device comprising:
-
a reduced-key keypad comprising a plurality of multi-character keys; and a processor cooperating with said reduced-key keypad for predicting characters input therein and being configured to populate a multi-level rule tree comprising a plurality of rules, each rule configured to predict a character in a current input word, rules in a respective level configured to predict a numbered character position in the current input word, the multi-level rule tree comprising a plurality of branches connecting the plurality of rules together, each branch being associated with a respective two-character sequence in the current input word, an upper level rule configured to predict a first character in the respective two-character sequence, and a lower level rule configured to predict a second character in the respective two-character sequence, each rule having a counter value representing a number of correct predictions, for each previously correct prediction of the current input word using a respective rule, increment the corresponding counter value in the multi-level rule tree to weight the respective rule, delete from the multi-level rule tree at least one rule having a weight value less than a threshold value, and weight subsequent predictions of other input characters based upon the counter values in the multi-level rule tree. - View Dependent Claims (13, 14, 15, 16, 17, 18, 19, 20, 21, 22)
-
Specification