A survey on location estimation techniques for events detected in Twitter

A spatial‐temporal‐semantic approach for detecting local events using geo‐social media data

“…In fact, the way Twitter data are collected acts as a major indicator for the geo‐tagged percentage. It illustrates that a geo‐bounding box‐based query method contributes to a higher percentage since crawled tweets meet the geo‐spatial requirement for GPS coordinates (Ozdikis, Oğuztüzün, & Karagoz, ). Among all geo‐tagged tweets collected through the Twitter Streaming API with a defined geospatial bounding box, as many as 30.11 and 27.49% of them originated from Asia and North America, respectively (Morstatter, Pfeffer, Liu, & Carley, ).…”

City-level Geolocation of Tweets for Real-time Visual Analytics

“…Geolocation inference at the event level estimates the location of events mentioned in text. This level of inference predominantly relies on geoparsing -the process of identifying geolocations in text and disambiguating between multiple toponym referencesand has been studied extensively [2,7,15]. Recent studies integrated Twitter metadata and named entity recognition algorithms into geoparsing approaches and obtained high accuracy percentages over 90% [3,7].…”

Quantifying uncertainty in population weighting of Twitter analyses for urban risk assessment

“…For HADR and other event-related efforts to use social media for crisis management, finer-grained geolocation techniques (i.e., below the city level) are not only more desirable statistically but can be vital to accomplishing operational mission goals. Ozdikis et al (2017) published a review studies focused on of targeted event types and granularity of estimated locations, the majority of which are below the city level (Table 1, p. 299).…”