Exploring Amharic Sentiment Analysis from Social Media Texts: Building Annotation Tools and Classification Models

Proceedings of the 28th International Conference on Computational Linguistics, pages 1048–1060 Barcelona, Spain (Online), December 8-13, 2020 1048 Exploring Amharic Sentiment Analysis from Social Media Texts: Building Annotation Tools and Classification Models Seid Muhie Yimam1, Hizkiel Mitiku Alemayehu3, Abinew Ali Ayele1,2 and Chris Biemann1 Universität Hamburg, Germany1, Bahir Dar University, Ethiopia2, Università di Bologna3, Italy yimam@informatik.uni-hamburg.de Abstract This paper presents the study of sentiment analysis for Amharic social media texts. As the num- ber of social media users is ever-increasing, social media platforms would like to understand the latent meaning and sentiments of a text to enhance decision-making procedures. However, low- resource languages such as Amharic have received less attention due to several reasons such as lack of well-annotated datasets, unavailability of computing resources, and fewer or no expert researchers in the area. This research addresses three main research questions. We first explore the suitability of existing tools for the sentiment analysis task. Annotation tools are scarce to support large-scale annotation tasks in Amharic. Also, the existing crowdsourcing platforms do not support Amharic text annotation. Hence, we build a social-network-friendly annotation tool called ‘ASAB’ using the Telegram bot. We collect 9.4k tweets, where each tweet is annotated by three Telegram users. Moreover, we explore the suitability of machine learning approaches for Amharic sentiment analysis. The FLAIR deep learning text classifier, based on network embed- dings that are computed from a distributional thesaurus, outperforms other supervised classifiers. We further investigate the challenges in building a sentiment analysis system for Amharic and we found that the widespread usage of sarcasm and figurative speech are the main issues in dealing with the problem. To advance the sentiment analysis research in Amharic and other related low- resource languages, we release the dataset, the annotation tool, source code, and models publicly under a permissive. 1 Introduction Sentiment analysis is the task of detecting the orientation of someone’s opinion and analyzing the emo- tions, feelings, and attitudes of a speaker or a writer in a piece of information concerning a certain sit- uation, object, or event (Pandey and Govilkar, 2015). The most widely adopted approach in sentiment analysis to explore opinions is by employing very large datasets that target products and services, politi- cal, economical, social, and cultural feelings (Kauffmann et al., 2019; Caetano et al., 2018; Lennox et al., 2020). Understanding the sentiment of text content helps governments, organizations, and institutions to make correct, timely, and economical decisions (De Souza Bermejo et al., 2019). Sentiment analysis has been researched intensively for resource-rich languages such as English and German (Liu, 2012; Feldman, 2013; Tymann et al., 2019; Akhtar et al., 2016; D’Andrea et al., 2015; Wojatzki et al., 2017). However, existing models and approaches for most resource-rich languages can not easily be adapted to Amharic due to context variations in language, culture, and technology, especially for social media communication (Gangula and Mamidi, 2018). The works by Gezmu et al. (2018) and Abate and Assabie (2014) indicate that natural language processing (NLP) components, such as part of …

Content truncated to 500 words