Major Advisor: Getachew Mamo(Ph.D.) Abstract The processes of developing and accessing arguments in the context of a discussion, dialogue or conversation are known as argumentation. Manually extracting argument relations in today's information-overloaded world is time-consuming, knowledge-intensive and prone to prejudice. Argument mining's overall purpose is to automatically identify and extract arguments as well as their relationships from huge unstructured data that may be used by a reasoning engine or computational model. Individual and group decisions, text summarization, corporate and governmental analyses and so on all benefit from it. Many attempts have been made to forecast argument relations for various languages. The majority of the research however focused on English and other European languages. Argument mining has two key subtasks: argument component and relation categorization (predictions). The study is conducted by supervised machine learning to design and execute argument relation prediction for the Afaan Oromo language in this study. Preprocessing is a component of the proposed system that removes superfluous data and prepares the input for the next component of the learning algorithm. Discourse marker features are generated using feature generation. Finally, for argument relation prediction tasks, feature extraction was employed to extract specific features. To test the argument relation prediction the research task consists of five tests with three classes and two classes utilizing supervised learning methods and 840 argumentative statements. From the result the acquired greatest weighted averages of F-score 85 percent and 81 percent in experiments in three and two classes respectively using Multi-Layer Perceptron (MLP) and Perceptron from the experiments. According to the findings, utilizing three-class predicting for MLP classifiers is a better learning classifier for Afaan Oromo argument relation prediction task. Keywords: Argument, Argument mining, Argument Relation Prediction, Word2vec, Discourse Marker, Proposition Similarity.