Only Thumbs Down? Samsung Galaxy Note7 Fiasco Twitter Sentiment Analysis Using Hybrid Method

With the almost ubiquitous access on the web, people use microblogs like Twitter, Facebook and Weibo to express their opinions on a wide variety of topics such as products, services, events, organizations, etc. Sentiment analysis on tweets has become a rapid and effective way of gauging opinion for business marketing or social studies. Unlike large opinionated corpora such as products reviews, tweets have unique characteristics that require special treatment to analyze the sentiment they convey. In this paper, we use the hybrid method combining the lexicon based and machine learning-based methods to perform sentiment analysis on tweets in the aftermath of Samsung Galaxy Note7 fiasco. First, we apply the lexicon-based approach todetermine the semantic orientations of opinions expressed in the tweets. This method gives high precision but low recall. To solve this problem, additional opinionated tweets are identified among tweets previously classified as neutral by handling words that are context dependent. We use Pearson's chi-square test to identify opinion words which are not in the lexicon and use them to train a support vector machines (SVMs) classifier to assign polarities to other additional tweets. This method is effective since it does not involve any manual labeling of tweets and has the ability to automatically adapt to new fashions in language, neologisms,and trends found in tweets. In this paper, we found that even though the Galaxy Note7 was removed from the market and killed completely by Samsung due to its battery catching fire, more customers still rated its feature more positive and labeled it as the top Android phone of its time.