Recognizing misinformation using geolocation: Spatial examination of Information Inferred from Tweets meta-fields

This is the authors' abstract. We don't add key points for this paper.

In recent years, Twitter has surged as a formidable communication platform, hosting over 500 million users worldwide. Despite a mere fraction of tweets (1–3%) containing direct geolocation, the abundance of metadata fields, including user-generated content, holds promise for inferring crucial spatial information. This study, aligning with principles of recognizing fake news and bolstering science credibility and public awareness, utilized GeoNames and OpenStreetMap datasets alongside a curated dataset of Tweets. Leveraging these resources, it aimed to infer tweet locations from text-based meta-fields and the user’s profile of location and description, emphasizing the verification of inferred locations' reliability amidst the prevalence of misinformation. Analysis revealed a strong correlation between distances derived from textual content and user-provided location meta-fields, indicative of potential links to actual device locations and phenomena described within tweets. Emphasizing the critical principles of recognizing misinformation and verifying geolocated data in tweets, this study emphasizes the importance of fostering science credibility and public awareness. It highlights the potential for accurately inferring device locations and validating information shared in tweets thus fostering the ability to detect fake news and promoting accurate scientific information dissemination for broader public awareness.