Qpen Source Spell Checker For Bukusu Dialect

Abstract

This project describes the development of an open source spellchecker for Bukusu language using Hunspell language tools. It examines Bukusu morphology, highlighting the inflection of various parts of speech in the Bukusu language including verbs, nouns and pronouns. The project focuses on the definition two major Hunspell files in the realization of the spell checker, i.e. the affix file (.aff) and the dictionary file (.dic). The dictionary file contains the Bukusu lexicon, whereas the affix file contains stemming rules that facilitate word suggestion using character prevalence and replacement procedures. The functional system is aimed at being adopted in major open-source products such as Open Office, Google Chrome, Firefox and Thunderbird. The open-source platform Used in the development of the spell checker will also allow for future improvements on the system. The developed spell checker realizes an acceptable performance with a precision of 79.3%, a recall of 77% and an accuracy of 68.9%