Dynamic sentiment sensing of cities with social media data

A lot of people share their living or travelling experiences about cities by writing posts on social media. Such posts carry multi-dimensional information about the characteristics of cities from the public’s perspective. This paper aims at applying text mining technology to automatically extract city images, which are known as how observers perceive the status of the city, from these social media texts.

This paper proposes a data processing pipeline for automatic city image extraction and applies sentiment analysis, timing analysis and contrastive analysis in a case study on Wuhan, a central China megacity. Specifically, the city image constructed with social media text and the expected policy outcomes by the government are compared.

Results reveal gaps between the public’s impression and the strategic goals of the government in traffic and environment.

This study contributes a novel approach to assess government performance by complementary data from social media. This case study implies the value of social media-based city image in the identification of gaps for the optimization of government performance.