Building Machine Learning Based Amharic Language Intent Classification Model

Abstract

Natural language processing is a science that explores how systems read and interpret people's language. In recent years, social networks have become extremely popular. Via various social media users are far more likely to express their everyday life, ideas or intentions. Intent analysis is an approach used to analyze user generated contents to a way that is important for decision making. This research benefits both company's or service provider and customer in terms of making a wise and effective decision, Major Benefits of Social Media for Businesses Improved customer insights, Better customer service. The experiments are conducted on data that collected from You-tube API during the simplicity of data scraping and filtering features. The aim of these research analyzing Amharic text and extract intentions behind a huge text data. And classify tokens into five classes (positive, negative, suggestion, wish, and question). In addition, investigate the impact of noise input data which misspelled and extra white space could affect proposed model.