Enhancing Social Media Event Geolocation with Geo-Tag and IP Address

Leveraging social media content for event geolocation is crucial for crisis management and resource allocation. Traditional methods for event geolocation predominantly rely on geotags or toponyms extracted from post text. However, these methods often leverage only a single location feature and neglect the varying importance of diverse locational information, which ultimately limits the precision of event geolocation. This paper proposed a novel event geolocation model, named Event Geolocation with Geotags and IP address (EGGI). This model integrates geotags, toponyms in text, and IP address from user profiles to achieve more accurate event geolocation. Initially, the model analyzes the hierarchical structure of toponyms within the event cluster and identifies the highly frequent city-level locations as the coarse-grained locations for the event cluster. Subsequently, it compares this location with the geotags/IP geolocations of the posts to filter out high-value posts. Toponyms extracted from these high-value posts are assigned higher weights. Finally, by geocoding these weighted toponyms and calculating their weighted centroid, the geolocation result of the event is obtained. Experimental results demonstrate that the proposed EGGI model exhibits significant performance advantages in the task of event geolocation.