Place Pulse : measuring the collaborative image of the city

[摘要] This thesis presents Place Pulse, a tool capable of conducting large crowdsourced visual preference surveys. The data collected with Place Pulse was used to create quantitative measures of the perceptions people hold of urban environments. From this data, novel algorithms identified locations associated with positive and negative perceptions of safety, social class and uniqueness. The high throughput of the tool addressed two important methodological questions: the number of responses required to obtain robust results in a comparative study, and the number of images required to get a statistically significant evaluative map of a large city. In closing, the validated dataset was used to correlate perceptions of safety and social class to rates of both violent crime and high school graduation.

[发布机构] Massachusetts Institute of Technology