Context-Based Auto Text Completion System for Amharic Language
The use of computers and handheld devices leads to easiness in communication using Short Service Message (SMS), E-mailing, correcting misspelled words and checking the grammatical mistakes. Out of all those usages of computers machines, we select the spelling checker and stand to develop a context-based auto text correction system for Amharic language specifically to correct misspelling on E-mailing and correcting the grammar mistakes as well. As we know, there are different data entry mechanisms to insert Amharic text on the computer machine as well as a mobile device, such as a keyboard, soft keys, speech etc. Data entry technique can be inserted with the support of Auto text correction (predictive) or non-predictive. Therefore by using Predictive Partial Match (PPM), Support Vector Machine (SVM), and statistical model, we develop context-based auto text correction for Amharic language. Since the system is developed on context-based, we have collected 750,000 Amharic sentences to identify the category of each word part of speech tagging and we have extracted 395,464 unique words from those sentences to build the Amharic dictionary as well as to calculate the frequency occurrences of each word. Finally, the results show a 14% improvement from traditional frequency-based Amharic word prediction system.
Keywords - Context-Based, POS, PPM, SVM.