Developing resources for sentiment analysis of informal Arabic text in social media

Natural Language Processing (NLP) applications such as text categorization, machine translation, sentiment analysis, etc., need annotated corpora and lexicons to check quality and performance. This paper describes the development of resources for sentiment analysis specifically for Arabic text in social media. A distinctive feature of the corpora and lexicons developed are that they are determined from informal Arabic that does not conform to grammatical or spelling standards. We refer to Arabic social media content of this sort as Dialectal Arabic (DA) - informal Arabic originating from and potentially mixing a range of different individual dialects. The paper describes the process adopted for developing corpora and sentiment lexicons for sentiment analysis within different social media and their resulting characteristics. The addition to providing useful NLP data sets for Dialectal Arabic the work also contributes to understanding the approach to developing corpora and lexicons.