An N-gram based model for predicting of word-formation in Assamese language

摘要

Word prediction is a technique which try to suggest the word by observing the previous input letters or words in any text editor. At present there is no such software or tool in Assamese which can predict the future word(s) of a sentence. This method helps the people who are not very much expert in typing and this research aims to reduce the gap between the people who are very much expert in typing with the people who are differently abled or the people who are not the daily users of the computer system. To predict the words, N-gram based models like unigram, bigram, trigram and quadrigram are used in this work. After doing two different level of experiments and testing maximum keystrokes saving (KS) 74.04% and 48.28% are found for preconfigured data-set and user-input data respectively. The results indicate a significant level of improvement towards sentence completion with the help of prediction method in Assamese language.