Automatic Tweet Mention Recommendation in X for Reporting Civic Issues - Case Study Based on Mumbai City, India

Aayush Patel, Chaitya Dobariya, Dayanand Ambawade, Dhawal Thakkar, P. Balamurugan

Abstract: X (Musk, 2022), formerly known as Twitter, has emerged as a popular venue for people to express their opinions and concerns about several topics and issues. This forum is used by millions of daily active users (citizens as well as government officials and individuals from private sectors) for public communication. However, the usefulness of this medium in resolving public problems is hampered by difficulties such as improper tagging of competent entities, which results in delayed or inefficient responses to issues. This study proposes a unique way to resolve problems on X that utilizes Natural Language Processing (NLP) techniques. The paper focuses on increasing the accuracy of recommending references to relevant authorities or departments, often denoted by a mention of the form @entityname, allowing for faster and more targeted responses to public concerns. We consider a case study in Mumbai, India, evaluating tweets on common urban challenges such as water scarcity, traffic congestion, and road upkeep, among others. We develop an NLP-based model to assess tweet content and recommend suitable mentions, ensuring that tweets reach the proper authorities effectively. We have further created a browser plugin based on our algorithm, which will suggest appropriate mention to the user based on the tweet in real-time. Furthermore, a web interface is designed to assist authorities in receiving actionable responses via alerts, therefore boosting clarity regarding the type, location and timing of reported concerns.