Phonetic-Based Microtext Normalization for Twitter Sentiment Analysis

摘要: The proliferation of Web 2.0 technologies and the increasing use computer-mediated communication resulted in a new form written text, termed microtext. This poses challenges to natural language processing tools which are usually designed for well-written text. paper proposes phonetic-based framework normalizing microtext plain English and, hence, improve classification accuracy sentiment analysis. Results demonstrated that there is high (>0.8) similarity index between tweets normalized by our model human annotators 85.31% cases, an increase >4% terms polarity detection after normalization.