The implementation of statistical word prediction algorithm in Tigrigna Word Prediction system text entry on android phone

Abstract

Text entry techniques are used to improve the interaction between the user and the handheld devices like smart phone, PDA,tablets, among other mobile devices. Text entry in these devices could be either predictive or non-predictive. Word prediction is a natural language processing problem that tries to predict the accurate or most appropriate candidate word in a given context. The conventional prediction systems use word frequency lists to complete intended words which the user wants to write.

Few researchers address Word Prediction Model for different language using a lexicon. However, as far as the researchers of this study are concerned no work is done to address word prediction system for Tigrigna text entry method. In this work,the researchers proposed for Tigrigna language a Tigrigna Word Prediction System for text entry on Android phone. They also usedsimple design method to sort the complete lexicon into their frequency order.There were 430,379 distinct Tigrigna words used to create the database and each word is sorted based on the frequency for predicting the word. The method is implemented and measured using 400 distinct words.

To test the validity of the word prediction model and the algorithm design, an experiment was conducted to measure the accuracy of the word prediction and keystroke saved. The result of Prediction accuracy was 91.6%; and the keystroke saved was 48.9%. Thisshowed that the method was effective in predicting the correct words, reduced the input burden of writing Tigrigna words on smart mobile phone by saving the keystroke.

The researchers based on findings, conclude, thatTigrigna Word Prediction System using Statistical Word Prediction Algorithm using Frequency for text entry is useful for Tigrigna language users to write Tigrigna texts on mobile phone.

Keywords: Algorithm, Tigrigna, Prediction, System, Text, Android. Phone