Natural Language Processing to forecast 2020 Chilean national plebiscite results: a Daoura project

Abstract

Cities around the world need to use advanced technology and foster innovation to provide better services to their citizens. One of the most evolving technological developments in recent years for analyzing the vast amount of information generated by billions of texts published by citizens in the Internet is Natural Language Processing (NLP), especially when used for text categorization, sentiment analysis and topic extraction. With the advent of NLP techniques, Artificial Intelligence has become applicable in different areas of science, business, and government. Although there have been advances in the application of Artificial Intelligence methods in societal problems, there is a strong lack of technologies for urban ser-vices management. This paper presents Daoura, a data intelligence platform for smart city management that transforms publications of citizens on the Internet into relevant insights, and its case on forecasting 2020 Chilean national plebiscite results for a new constitution, based on analyzing millions of citizens interactions in social media. The platform's projections reported an accuracy of 95.8% when compared to the final results, showing the importance of digital manifestations on understanding people's opinions and experiences in the society. We also discuss the significative impacts Daoura NLP skills can have on modernizing public management and creating environments that promote well-being and quality of life in cities.