International Journal of Computer Science Trends and Technology (IJCST) – Volume 4 Issue 3, May - Jun 2016 ISSN: 2347-8578 www.ijcstjournal.org Page 173 Sentiment Analysis of Arabic Tweets: With Special Reference Restaurant Tweets Mnahel Ahmed Ibrahim [1] College of Computer Science and Information Technology [1] Sudan University of Science and Technology Khartoum, Sudan Naomie Salim [2] Faculty of Computing [2] Universiti Teknologi Malaysia Johor Malaysia ABSTRACT Sentiment analysis is an emerging area, its use in industry and business sector is increasing due to provision of sentiment of the core information applications are requiring. The product manufacturers always require some method to know the future acceptance of their product in production. This is among the key knowledge to enhance the quality of the produc t and improve the underlined process. The research undertakes sentiment analysis of Arabic tweets extracted through Twitter microalgae. The methodology was divided in two phases. Analysis of text models for many combinations of text is done; few examples are stop word reduction, stemming, and n-gram. The other phase applied and analyzed results of classifiers like Naive Bayes, SVM and K-nearest neighbour. Arabic Sentiment Analysis is a challenging task because, applications developed for Arabic natural language processing focused only modern Arabic, while less focusing the various dialects in operation by different regions like North Africa, Egypt and Gulf States. Rapidminer is a tool used to conduct the simulation. Semantic analysis classifiers are developed to learn manually annotated data. This yielded eighty eight percent accuracy of cross validation. It is evident that picking preprocessing strategies for the reviews enhances the ability of the classifiers. In future the research aims to solve target identification of tweets and classify tweets with the help of patterns. Keywords: - Sentiment Analysis, opinion mining, Arabic, Twitter. I. INTRODUCTION In the last decade text classification and summarization becomes an attractive and interesting area for researchers and developers of data mining. There is an acute requirement for auto extraction and summarization of massive and significant information in this area. This information can be used by users and applications. Lot of research has been done to address the issue. The most popular technique is categorization of topic. In this technique documents are categorized on the basis of strategies represented by them [1], [2].Currently a technique has evolved to categorize the documents on the basis of opinions made by users or according to their sentiments[3]– [6].This means to classify documents as positive or negative based on associated data[7]. The main generator of this research area is social networks and bloggers. These online networks are driven by people’s choices and accepted or rejected by their say. The websites are now rich with different commercial aspects, knowledge stores, entertainment sources, books and articles and much more. The study is coming up with initial processing and feature extraction techniques for Arabic tweets. Precision and accuracy for Sentiment Analysis (SA) will be measured too. Classifiers like Naive Bayes, K-Nearest neighbor (K- NN) and SVM are used for validation. Rapidminer is used for experimentation [30]. Arabic is the language of the Holy Quran used by almost one billion Muslims across the world, different dialects are spoken by nearly 200 million people. Arabic is the main language for twenty two Arab league members, and official language of 3 countries [8]. The Arabic language has major differences from most popular languages like English and Chinese. Arabic has many grammatical forms, varieties of word synonyms, and different word meanings that vary depending on many factors like word order [9]. RESEARCH ARTICLE OPEN ACCESS