3rd International Conference on Smart Sustainable City and Big Data 27-28 July 2015, Shanghai, China
This research uses a Big Data methodology for under-standing geographic social-cultural indicator... more This research uses a Big Data methodology for under-standing geographic social-cultural indicators in Smart Cities that can be generalised to other topics, by investi-gating factors influencing sentiments of city dwellers via large-scale social media. We adopted an important indica-tor as a proof of concept reported here – alcohol consump-tion and factors influencing sentiments of world-wide cities. Big Data methodological approaches were used for collecting, pre-processing, analysing and mapping senti-ments over time. Geo-referencing was used for mapping 24-hour activities in relation to density and volume of activities, together with sentiment analysis, and multino-mial logistic regression. The results demonstrate the feasi-bility of our Big Data approach in Smart City sentiment monitoring. Based on a dataset of more than 369,000 tweets, our work shows that for emotionally charged tweets, the count of followers, location, alcohol strength and the sentiment of self-description are significant influ-encing variables. The research has definite implications in Smart City governance of public health, safety, and the public monitoring and mapping of hotspot areas at differ-ent time of day in real-time, and therefore the potential to predict the spread of behaviours that could impact on pub-lic safety. The generic methodology could be adopted for sentiments of variable topics in heterogeneous datasets.
3rd International Conference on Smart Sustainable City and Big Data 27-28 July 2015, Shanghai, China
This research uses a Big Data methodology for under-standing geographic social-cultural indicator... more This research uses a Big Data methodology for under-standing geographic social-cultural indicators in Smart Cities that can be generalised to other topics, by investi-gating factors influencing sentiments of city dwellers via large-scale social media. We adopted an important indica-tor as a proof of concept reported here – alcohol consump-tion and factors influencing sentiments of world-wide cities. Big Data methodological approaches were used for collecting, pre-processing, analysing and mapping senti-ments over time. Geo-referencing was used for mapping 24-hour activities in relation to density and volume of activities, together with sentiment analysis, and multino-mial logistic regression. The results demonstrate the feasi-bility of our Big Data approach in Smart City sentiment monitoring. Based on a dataset of more than 369,000 tweets, our work shows that for emotionally charged tweets, the count of followers, location, alcohol strength and the sentiment of self-description are significant influ-encing variables. The research has definite implications in Smart City governance of public health, safety, and the public monitoring and mapping of hotspot areas at differ-ent time of day in real-time, and therefore the potential to predict the spread of behaviours that could impact on pub-lic safety. The generic methodology could be adopted for sentiments of variable topics in heterogeneous datasets.
Uploads
Publications by Mengdi Li