Sentiment-Based Machine Learning Approach for Mapping Citizen Problems

Computer science and Engineering

In today's digital era, social media platforms have become a key medium for citizens to express concerns and grievances. This project proposes a real-time system for detecting and classifying citizen problems from social media posts using a sentiment-based machine learning approach. Due to the massive volume and dynamic nature of user-generated content, manual analysis is impractical. To address this, we leverage machine learning and natural language processing (NLP) techniques to filter spam, detect relevant problems, and classify issues in real time. The system follows an end-to-end pipeline involving data extraction, preprocessing, problem detection, sentiment analysis, and location mapping. Social media APIs and web scraping methods gather real-time data, which is refined through text cleaning, tokenization, and stemming. Supervised learning models detect and classify problems, while NLP techniques like named entity recognition and geoparsing extract location information to map citizen concerns geographically. The proposed system demonstrates the potential to automate public grievance monitoring and provide actionable insights. By combining sentiment analysis, machine learning, and geolocation, it supports faster response to emerging issues, improves public services, and enhances citizen engagement through data-driven decision-making.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.