Abstract:
This paper presents Afan Oromo semantics which is identifying the words semantically related. Semantic is
one of the critical application in natural languages, hence it is a fundamental problem for many natural
language technology applications. The aim of this work is to develop sense disambiguation which finds the
sense of words based on surrounding contexts. Hence, this study used unsupervised approach that exploits
sense in a corpus which is not labelled. The idea behind the approach is to overcome the problem of scarcity
of training data. The context of a given word is captured using term co-occurrences within a defined window
size of words. The similar contexts of target words are computed using vector space model and then
clustered. From total clustering, each cluster representing a unique sense. Most of the target words have
more than three senses. The result argued that the system yields an accuracy of 85% which was encouraging
result. Therefore, for Afan Oromo semantic has come to the conclusion that the sense of words is closely
connected to the statistics of word usage. Further study using different approaches that extend this work are
needed for a better performance.
