Sentence prediction system for Afan Oromo text

Abstract

Data entry is a core aspect of human computer interaction. Sentence pre diction is one of data entry systems to a computer and other hand held electronics device. It is a process of guessing the phrases or words which are likely to follow in a given text segment by displaying a list of the most probable sentences that could appear in that position. Sentence prediction assists physically disabled individuals who have typing difficulties, speed up typing speed by decreasing keystrokes, helps in spelling and error detection and it also helps in speech recognition and hand writing recognition. In this study, Sentence prediction model is designed and developed for Afan Oromo. We used RNN-LSTM for deep learning algorithm. Initially, the training cor pus and user inputs are tokenized and analyzed. Subsequently, RNN-LSTM model is built on pre processed and cleaned data by using training corpus. After all the model is tested by input provided from the users. The developed model is evaluated based on Keystroke Saving (KSS) performance evaluation metrics.According the evaluation result 18.75% KSS is achieved based on the average of KSS of four testers.Therefore,Recursive neural network has good potential on sentence prediction for Afaan Oromo.