The introduction of web 2.0, has created the unprecedented eruptions of users generated contents. The eruptions of ample of reviews have created an opportunity for the new research dimension called opinion mining. Due to the vast amount of online available reviews, it's very tough for users to read and produce a useful summary of the reviews provided with different aspects. Despite their usefulness if they are properly structured, analyzed and presented to the users, these reviews are left unused properly. Most of the popular online opinion mining system caters only for the most resourced languages (i.e. English, Arabic) but no research has been conducted so far concerning the opinion mining of under-resourced Afaan Oromo language. This paper, proposed an Enhanced Vector Space Model (EVSM) algorithm, for opinion mining of Afaan Oromo reviews under online news sites. The major novelty of this paper is the use of machine learning, for development of the opinion mining model of Afaan Oromo reviews and its self-annotation ability. The experiment confirms that the proposed approach has shown a significant improvement in the output of the developed model. We put forward working on the efficiency of the developed model.

Keywords: Afaan Oromo; aspect based opinion mining; enhanced vector space model; news Sites; self-annotation