ADAL System: Aspect Detection for Arabic Language

Sentiment analysis can be done at different levels of granularity: document, sentence, and aspect. In our case, we are interested in the aspect which presents the finest level of granularity. This level is named Aspect Based Sentiment Analysis. In fact, Aspect Based Sentiment Analysis (ABSA) requires two primordial steps: (i) extract entity aspects and (ii) determine the sentiments from all the aspects. Aspect extraction is an important step for the ABSA. It aims at detecting all the existing aspects in a sentence. The extraction of these aspects is complicated considering the presence of several challenges especially when the Aspect extraction is done in the Arabic language. In this paper, we propose a supervised system ADAL for aspects detection in the Arabic language. The obtained results indicate that our proposed method outperforms previous works to achieve 96% in terms of f-measure when applied to the same dataset provided by The International Workshop on Semantic Evaluation 2016 (SemEval-2016).