Geolocation prediction in social media data using text analysis: A review

Abstract

Geolocation information from social media data is essential for conducting geolocation-based analyzes such as traffic analysis and tourism analysis. However, geolocation information on social media data is still very limited. Only about 0.87% to 3% of data are geotagged data. Geolocation Prediction (GP) becomes a solution to overcome the problem. There are various approach to conduct Geolocation Prediction and each approach may give different result of location. The selection of the Geolocation Prediction approach then become important. Selected approach must be suitable for the needs of the analysis conducted. This paper focuses on reviewing geolocation prediction approaches based on text analysis in social media data. The review result shows that geolocation prediction approaches can be categorized into two categories called Content-based Geolocation Prediction and User-profiling-based Geolocation Prediction. This review further concludes that Content-based Geolocation Prediction is suitable for addressing geotagged data limitations in Location-specific Analysis because the location prediction results are specific to place-level. While combination approach is suitable to overcome the problem of geotagged data limitations on Location-distribution Analysis because it produces predictions of location at higher levels such as city-level, province-level, and country-level.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 28 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.