TWITTER MINING: A Predictive Model for Public Opinion trained on Citizen Engagement in 100 Smart-Cities.

I am investigating civic technologies and decision-making in smart-cities, in order to enlighten the urban professionals of all field on the state of the art of citizen-led initiatives for future planning. In this article, I keep developing my Artificial Intelligence for Smart-City by adding a predictive model for lexical content.

To do so, I have been collecting 110,862 tweets (more than 19 Millions of words) over the 109 most advanced Smart-Cities worldwide, ranked by the IMD Smart City Index 2020. The three case studies of my doctorate degree’s thesis are: Taipei (Taiwan), Tel Aviv (Israel) and Tallinn (Estonia). They are three of the 109 samples, on which I collected significantly more tweets than in other cities, riding up respectively to 2013 and 2012, which represents almost all the tweets that have been published on these cities since the creation of the social network.

I am using this lexical content to extract numerical attributes with Natural Language Processing (NLP) and Sentiment Analysis techniques, and other calculation on words such as the weight of specific Bags-of-Words (BoWs) for urban studies fields. My model is built with the programming language Python, whose code and the initial raw data file, are available on my Github.

=> To mention this article: “TWITTER MINING: A Predictive Model for Public Opinion trained on Citizen Engagement in 100 Smart-Cities.” Julien…