NIJ Innovations in Measuring Community Perceptions Challenge Winner: Community PoliSense

This entry proposes to use data from social media (X, formerly Twitter), administrative sources (voter registration), and Census to uncover community perceptions at microgeographies. Twitter names and self-reported locations would be matched with voter registration street address data to obtain individual-level demographics. This sample of geolocated Twitter users can then be enriched with Census block group information. Tweets and shared articles are processed with text classifiers to add tags, which would be used to train another natural language processing (NLP) model to identify the sentiment of tweets and calculate user-level sentiment scores measuring perceptions for each of the five constructs. This replicable methodology represents an innovative approach to identifying and measuring community perceptions of public safety-related constructs at microgeographies.

Team members included Annie Chen, Cecilia Low-Weiner, Osama Qureshi, and Michael A. Keith, Jr..

This entry won first prize in the category 2 overall competition.