Abstract
Recently, the use of word embeddings has become one of the most significant advancements in natural language processing (NLP). In this paper, we compared two word embedding models for aspect-based sentiment analysis (ABSA) of Arabic tweets. The ABSA problem was formulated as a two step process of aspect detection followed by sentiment polarity classification of the detected aspects. The compared embeddings models include fastText Arabic Wikipedia and AraVec-Web, both available as pre-trained models. Our corpus consisted of 5K airline service related tweets in Arabic, manually labeled for ABSA with imbalanced aspect categories. For classification, we used a support vector machine classifier for both, aspect detection, and sentiment polarity classification. Our results indicated that fastText Arabic Wikipedia word embeddings performed slightly better than AraVec-Web.