Adaptive and Optimization Predictive Text Entry for Short Message Service (Sms)

The ability to enter text on phones is crucial for utilizing the Short Message Service (SMS) of digital cellular networks. Standard 12-button phone keypads are not naturally suited for text entry as each button on them encodes three or more alphabetic letters. Various techniques have been proposed and implemented to improve text entry tasks. One such technique is predictive text entry. In this paper we describe the development of a generic mobile phone predictive text application. We have tried it on some local Kenyan languages. It is implemented as a word weighting system that is based on the sigmoid function. It initially relies on the natural word frequency of a given language. We also present an adaptive learning model for improving text entry speed that incorporates a specific user’s word usage habits. Our results indicate an improvement in the performance of the developed system as evaluated and compared to other current predictive text techniques.