Development of an English to Yoruba machine translation system using direct based approach,

ABSTRACT

Language is the basic medium of communication and it is ultimately important because it is

the primary means through which humans have the ability to communicate and interact with

one another. Translation has been made very important so as to bridge the gap of information

inequality. A lot of countries and organizations rely on human translators to disseminate

information but human translators are limited in terms of speed, insecurity and high cost,

therefore the demand for faster and cheaper means of translation. Hence, an English to Yoruba

statistical machine translator was developed in this research.

The project was developed using python programming language and PyQt designer was used

to design the graphic user interface. The direct method of translation was used to create a

database of 25,000 words, which were gotten from locally spoken words and dictionaries. The

system was evaluated using Mean opinion score (average) and tested using three different

datasets.

The results from evaluation of the system based on percentage accuracy obtained from three

different datasets are: 92.6%, 87.4%, 89.1% for dataset 1, 2 and 3 respectively. Thus, the

percentage accuracy of the developed system was found to be 89.7% and was derived from the

average percentage accuracy of the different datasets.

In conclusion, an English to Yoruba Translator was developed in this research to overcome the

shortcomings of human translators and aid commercial activities. It is recommended that

government should invest in machine translators to aid socio-economic growth of Yorubaspeaking

states.