Towards a Scalable Geoparsing Approach for the Web

Kurzfassung

The ongoing surge in web data generation and storage, coupled with embedded geographic information, holds immense potential for enhancing search applications across diverse domains. However, extracting geographic information for further enhancement of web search remains inadequately explored. This paper addresses a critical gap in the realm of geographic information extraction from web data, emphasizing the absence of unified pipelines for processing such information. In response to this void, we present a pipeline specifically tailored for web data. Furthermore, our contribution extends beyond the development of the pipeline itself to include a comparative analysis of various gazetteer-based geotagging methods in terms of accuracy and scalability along with a sizable corpora of location annotated web documents.