An Algerian Corpus and an Annotation Platform for Opinion and Emotion Analysis

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 1202–1210 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 1202 An Algerian Corpus and an Annotation Platform for Opinion and Emotion Analysis Leila Moudjari, Karima Akli-Astouati, Farah Benamara RIIMA Laboratory USTHB Bab Ezzouar 16111, Algiers Algeria IRIT Universit´e de Toulouse, 118 Route de Narbonne, l.moudj11@gmail.com/lmoudjari@usthb.dz, kakli@usthb.dz, farah.benamara@irit.fr Abstract In this paper, we address the lack of resources for opinion and emotion analysis related to North African dialects, targeting Algerian dialect. We present TWIFIL (TWItter proFILing) a collaborative annotation platform for crowdsourcing annotation of tweets at different levels of granularity. The plateform allowed the creation of the largest Algerian dialect dataset annotated for both sentiment (9,000 tweets), emotion (about 5,000 tweets) and extra-linguistic information including author profiling (age and gender). The annotation resulted also in the creation of the largest Algerien dialect subjectivity lexicon of about 9,000 entries which can constitute a valuable resources for the development of future NLP applications for Algerian dialect. To test the validity of the dataset, a set of deep learning experiments were conducted to classify a given tweet as positive, negative or neutral. We discuss our results and provide an error analysis to better identify classification errors. Keywords: Crowdsourcing annotation platform, Algerian dialect, sentiment analysis, emotion detection, subjectivity lexicon 1. Introduction Currently, there are more than 4 billion Internet users worldwide. More than 50% of the North African popula- tion has access to the Internet. The region has also seen a growth of more than 17% in the number of social media users compared to 2017 1. In Algeria, more than 50% of the population are registered users on different social platforms and around 46% of them use mobile devises for such activ- ity 2. These numbers represent a growth of 17% in social media use and more than a 19% growth in the use of mobile devises for such platforms. Twitter users in Algeria reached 8.73% in August 2019 compared to August 2018 (2.96%). The number has almost tripled over a year, making Twitter the third most used platform by active social media users 3. On social media platforms, 76% of users express their sen- timents by clicking corresponding buttons when available, such as ”Like”, ”Dislike”. Around 50% expresses views or sentiments using ”emoticons”, ”emojis” or ”smileys”. Across the Arab region, more than 30% of the users use Arabic script and 26% uses Latin script (mostly English and French) and about 15% combine both (Salem, Feb 5 2017). Compared to other Arabic dialects, the North African dialects have other peculiarities, as several lan- guages are used in everyday conversations. For example, the expression ”Nro7o ensemble?” is a combination of the French word ”ensemble” meaning ”together”, and the Ara- bic word ”nro7o (� ��� ����)”, meaning ”we go together”. Sentiment analysis and emotion detection in Arabic have been widely studied (Baly et al., 2017; Al-Smadi et al., 2018; Abo et al., 2018). Most related work focus on …

Content truncated to 500 words