Metaphorical Expressions in Automatic Arabic Sentiment Analysis

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 4911–4916 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 4911 Metaphorical Expressions in Automatic Arabic Sentiment Analysis Israa Alsiyat1,2 and Scott Piao1 1School of Computing and Communications, Lancaster University, UK 2Northern Borders University, Saudi Arabia {i.alsiyat,s.piao@lancaster.ac.uk; israa.alsiyat@nbu.edu.sa} Abstract Over the recent years, Arabic language resources and NLP tools have been under rapid development. One of the important tasks for Arabic natural language processing is the sentiment analysis. While a significant improvement has been achieved in this research area, the existing computational models and tools still suffer from the lack of capability of dealing with Arabic metaphorical expressions. Metaphors have an important role in Arabic language due to its unique history and culture. Metaphors provide a linguistic mechanism for expressing ideas and notions that can be different from their surface form. Therefore, in order to efficiently identify true sentiment of Arabic language data, a computational model needs to be able to “read between lines”. In this paper, we examine the issue of metaphors in automatic Arabic sentiment analysis by carrying out an experiment, in which we observe the performance of a state-of-art Arabic sentiment tool on metaphors and analyse the result to gain a deeper insight into the issue. Our experiment evidently shows that metaphors have a significant impact on the performance of current Arabic sentiment tools, and hence it is an important task to develop Arabic language resources and computational models for Arabic metaphors. Keywords: Sentiment Analysis, Metaphor, Arabic NLP, Evaluation 1. Introduction With the increase of the social media and e-commerce activities in the Arabic-speaking world, Arabic online tools and services have been increasing rapidly. As a result, the Arabic-speaking users have access to an increasing number of Internet channels to express their thoughts and opinions towards commercial products or other topics. As the result, the amount of Arabic textual data for sentiment analysis (SA) available online has been growing substantially, and there is an urgent need for developing corpus resources, efficient algorithms and software tools for detecting and analysing sentiment information from the Arabic language data. Over the past years, a variety of Arabic language resources, algorithms and tools have been suggested and developed, such as (Elarnaoty et al., 2012) and (Rushdi‐Saleh et al., 2011), with various levels of success. However, automatically identifying sentiment information from the Arabic language data is highly challenging. One of the major reasons for the difficulty derives from rich Arabic metaphorical expressions. Metaphor is intuitively fundamental in our ordinary life of thinking, speaking and action (Lakoff and Johnson, 1980). The figurative expressions, which covers the most equivocal expressions, or metaphors, is a non-literal sub-language of an imaginative and embodies speech. A purpose of using such expressions is to hide true feelings in certain situations. Metaphor is an interesting cognitive linguistic as well as technical issue of the Natural Language Processing (NLP) for understanding and computing nuanced linguistic semantic mechanism. While metaphor is a part of …

Content truncated to 500 words