WORD SEQUENCE PREDICTION FOR AMHARIC LANGUAGE

Nuniyat Kifle, Ermias Abebe

Abstract: Word prediction is one of the techniques to text entry and assistive technology for people with disabilities like Dyslexia which is problem of reading and spelling. For developing countries such as Ethiopia this kind of problems are neglected and the language spoken within the country are under resourced. Therefore applying AI to these problems has a major contribution. Amharic is used by a large number of populations, however no significant work is done on the topic of word sequence prediction. In this study, Amharic word sequence prediction model is developed with statistical methods using Hidden Markov Model by incorporating detailed parts of speech tag, some morphological features and user profiling or adaptation. Evaluation of the model is performed using developed prototype and keystroke savings (KSS) as a metrics. According to our experiment, prediction results using a bi-gram with morphological features and detailed Parts of Speech tag model has higher KSS and performed better compared those without specified features. Therefore, statistical approach with detailed POS, morphological features like gender, number, and person with suggested root or stem words using voice, tense, aspect, affixes statistical information and grammatical agreement rules of the language has quite good potential on word sequence Prediction for Amharic language.

TL;DR: Amharic word sequence prediction model is developed with statistical methods using Hidden Markov Model by incorporating detailed parts of speech tag, some morphological features and user profiling or adaptation.