Regressing Location on Text for Probabilistic Geocoding

TL;DR

This paper introduces ELECTRo-map, a probabilistic model for geocoding text data, demonstrating improved uncertainty estimation and contextual understanding over existing systems, supported by a new evaluation dataset.

Contribution

The paper presents a novel end-to-end probabilistic geocoding model and a new dataset for evaluating geocoding performance.

Findings

ELECTRo-map outperforms existing open-source geocoding systems.

The model provides principled uncertainty estimates.

Leveraging context improves geocoding accuracy.

Abstract

Text data are an important source of detailed information about social and political events. Automated systems parse large volumes of text data to infer or extract structured information that describes actors, actions, dates, times, and locations. One of these sub-tasks is geocoding: predicting the geographic coordinates associated with events or locations described by a given text. We present an end-to-end probabilistic model for geocoding text data. Additionally, we collect a novel data set for evaluating the performance of geocoding systems. We compare the model-based solution, called ELECTRo-map, to the current state-of-the-art open source system for geocoding texts for event data. Finally, we discuss the benefits of end-to-end model-based geocoding, including principled uncertainty estimation and the ability of these models to leverage contextual information.