PRINT READER OUTPUT CORRECTION STUDY.
Final rept. Jun 66-Jun 67,
IBM FEDERAL SYSTEMS DIV GAITHERSBURG MD
Pagination or Media Count:
This study resulted in the development of a basic model for assisting an optical character recognition unit in deciding character identity by context dependent factors. A programmed method was designed to introduce controlled errors in a textual data base to simulate the characteristics of output from an optical character recognition device. Two error types rejects and best guesses, were created in the alphabetic words on this English textual material. A basic model was designed, implemented, and evaluated for effectively correcting these errors. Correction techniques used in this model are based, not on full dictionary lookup, but on n-gram occurrence lists common word dictionaries, environmental dictionaries, and character confusion tables. Author
- Information Science
- Computer Programming and Software
- Computer Hardware