The use of new technologies to support Public Administration. Sentiment analysis and the case of the app IO

TL;DR

This paper explores how sentiment analysis and emotion detection, using machine learning models, can identify issues in a public service app to improve citizen interaction and satisfaction.

Contribution

It introduces a methodology combining automated feedback scraping with machine learning for sentiment and emotion analysis in public administration apps.

Findings

Identified key factors causing negative reviews

Demonstrated effectiveness of sentiment analysis in public service feedback

Provided insights for improving app perception

Abstract

App IO is an app developed for the Italian PA. It is definitely useful for citizens to interact with the PA and to get services that were not digitized yet. Nevertheless, it was not perceived in a good way by the citizens and it has been criticized. As we wanted to find the root that caused all these bad reviews we scraped feedback from mobile app stores using custom-coded automated tools and - after that - we trained two machine learning models to perform both sentiment analysis and emotion detection to understand what caused the bad reviews.