PS-Radar: A High-Precision Geoparsing Solution for Real-Time Location Analysis
Abstract: PS-Radar is an advanced geoparsing solution utilizing OSM Nominatim to analyze text messages and generate a curated list of potential locations mentioned within. These results are then disambiguated and ranked based on their likelihood of accurately representing the target locations. Our system integrates data from OSM and other sources, enabling real-time map visualization, geofencing, area-based filtering, and alerting for monitoring social media message streams. Furthermore, we introduce the SonarChallenge dataset, comprising 1489 annotated messages containing location references. In our evaluation using the SonarChallenge dataset, our solution achieves an 88.49\% recall rate for identifying mentioned locations and a 92.51\% precision rate for accurately pinpointing the output locations.
Paper Type: Long
Research Area: Information Extraction
Research Area Keywords: event extraction, named entity recognition and relation extraction, open-domain QA
Contribution Types: NLP engineering experiment, Data resources
Languages Studied: English, Dutch
Submission Number: 509
Loading