oa An Efficient Model For Sentiment Classification Of Arabic Tweets On Mobiles

With the growth of social media and online blogs, people express their opinion and sentiment freely by providing product reviews, as well as comments about celebrities, and political and global events. These texts reflecting opinions are of great interest to companies and individuals who base their decisions and actions upon them. Hence, opinion mining on mobiles is capturing the interest of users and researchers across the world with the growth of available online data. Many techniques and applications have been developed for English while many other languages are still trying to catch up. In particular, there is an increased interest in easy access to Arabic opinion from mobiles. In fact, Arabic presents challenges similar to English for opinion mining, but also presents additional challenges due to its morphological complexity. Mobiles on the other hand present their own challenges due to limited energy, limited storage, and low computational capability. Since some of the state-of-the-art methods for opinion mining in English require the extraction of large numbers of features, and extensive computations, these methods are not feasible for real-time processing on mobile devices. In this work, we provide a solution to address the limitation of the mobile, and the required Arabic resources to derive opinion mining on mobiles. The method is based on matching stemmed tweets to our own developed Arabic sentiment lexicon (ArSenL). While there have been efforts towards building Arabic sentiment lexicons, they suffer from many deficiencies including limited size, unclear usability plan given Arabic's rich morphology, or non-availability publicly. ArSenL is the first publicly available large scale Standard Arabic sentiment lexicon (ArSenL) developed using a combination of English SentiWordnet (ESWN), Arabic WordNet, and the Standard Arabic Morphological Analyzer (SAMA). A public interface to browsing ArSenL is available at http://me-applications.com/test. The scores from the matched stems are then aggregated and processed through a decision tree for determining the polarity. The method was tested on a published set of Arabic tweets, and an average accuracy of 67% was achieved versus a 50% baseline. A mobile application was also developed to demonstrate the usability of the method. The application takes as input a topic of interest and retrieves the latest Arabic tweets related to this topic. It then displays the tweets superimposed with colors representing sentiment labels as positive, negative or neutral. The application also provides visual summaries of searched topics and a history showing how the sentiments for a certain topic has been evolving.