Lexicon Based Sentiment Comparison of iPhone and Android Tweets During the Iran-Iraq Earthquake

Due to the vast amount of digital information on social media, a platform like Twitter provides a rich corpus for data mining in general, and opinion mining and sentiment analysis in specific. Moreover, and since social data is mostly unstructured in nature, effective data processing and analytics techniques are needed. The literature on sentiment analysis outlines three main approaches: supervised, unsupervised, and hybrid; each having its applications and advantages. After the November 2017 earthquake that hit the Iran-Iraq border, 50000 tweets were collected over a six-day period. The data was cleaned to isolate and study iPhone and Android tweets pertaining to sentiment and emotion detection. In this case study, unsupervised classification utilizing a lexicon based approach was used. The sentiments of iPhone and Android tweets were classified based on the polarity of the sentiment (i.e. negative or positive) as well as emotions. The objective of this study was to quantify the observed difference in sentiment between the Android and iPhone tweets using the NRC Word-Emotion Association lexicon.