Reconhecimento de caracteres manuscritos utilizando regras de associação.

Abstract

In spite of the growing spread of information technologies based on electronic media, hand-printed information, such as forms, memoranda, letters, requirements, etc. is still abound. The conversion process of these stored forms to electronic media has been onerous and unproductive, mainly due to the involvement of unprepared personnel. This paper presents the results of a pioneer application of the data mining technology known as association rules for the automatic recognition of hand-printed characters. Besides analysing the results obtained with this technique, we also compare our system for automatic recognition of hand-printed characters with systems that use other tecnologies, as syntactic analysis and neural nets.