An Opinion Spread Prediction Model With Twitter Emotion Analysis During Algeria's Hirak. (14th September 2020)

• Record Type: Journal Article
• Authors: Drif, Ahlem; Hadjoudj, Khalil
• Is Part Of: Computer journal. Volume 64:Number 3(2021)
• Page Start: 358
• Page End: 368
• Publication Date: 2020-09-14
• Subjects: information propagation -- deep learning -- sentiment analysis -- features extraction -- human behaviour -- natural languages processing -- long short term memory network (LSTM) -- Hirak; Computers -- Periodicals; 005.1
• DOI: 10.1093/comjnl/bxaa117
• Languages: English
• ISSNs: 0010-4620

Abstract

Abstract: Social media is believed to have played a central role in the mobilization of Algerian citizens to peaceful protest against their country's corrupt regime. Since no one foresaw these protests (called 'The Revolution of Smiles' or 'The Hirak Movement'), this research conducted social media analysis to elicit vital insights about both the intensity of sentiment and the influence of social media on this unexpected instigation of political protest. This work built a deep learning model and analysed the influence of content, sentiment and user features on information spread. The model used the learning capability of a long short-term memory network to predict 'retweetability'. Experiments were conducted on two real-world datasets (Hirak and Brexit) collected from Twitter. User features were found to be a key element in the diffusion of information. The strongest feelings about event context actively influenced the spread of tweets. The Twitter emotion corpus was found to improve the predictive ability of the model developed in this study.