Automatic Construction of a Large-Scale Corpus for Geoparsing Using Wikipedia Hyperlinks

TL;DR

This work tackles the scarcity of large, general-domain geoparsing data by introducing WHLL, a method that leverages Wikipedia hyperlinks to annotate coordinates for multiple location expressions within 1.3 million articles. The WHLL corpus provides 14.7 million location expressions (1.6 million unique), with 45.6% ambiguous expressions and 9.9% recessive cases, enabling large-scale training and evaluation of geoparsing systems. Geocoding experiments comparing familiarity-based and dependency-based strategies demonstrate that incorporating linguistic context via Universal Dependencies improves disambiguation accuracy, though there remains room for further gains. Overall, WHLL offers a scalable, automatically constructed resource that supports training, evaluation, and improvement of geoparsing models across diverse geographic domains.

Abstract

Geoparsing is the task of estimating the latitude and longitude (coordinates) of location expressions in texts. Geoparsing must deal with the ambiguity of the expressions that indicate multiple locations with the same notation. For evaluating geoparsing systems, several corpora have been proposed in previous work. However, these corpora are small-scale and suffer from the coverage of location expressions on general domains. In this paper, we propose Wikipedia Hyperlink-based Location Linking (WHLL), a novel method to construct a large-scale corpus for geoparsing from Wikipedia articles. WHLL leverages hyperlinks in Wikipedia to annotate multiple location expressions with coordinates. With this method, we constructed the WHLL corpus, a new large-scale corpus for geoparsing. The WHLL corpus consists of 1.3M articles, each containing about 7.8 unique location expressions. 45.6% of location expressions are ambiguous and refer to more than one location with the same notation. In each article, location expressions of the article title and those hyperlinks to other articles are assigned with coordinates. By utilizing hyperlinks, we can accurately assign location expressions with coordinates even with ambiguous location expressions in the texts. Experimental results show that there remains room for improvement by disambiguating location expressions.

Figures (5)

• Figure 1: Coordinates annotation using hyperlinks. Location expressions with hyperlinks are annotated with the coordinates of the linked articles, and those equal to the article title are annotated with the coordinates of the article. A sample article of our corpus is shown in \ref{['fig-sample']}.
• Figure 2: A sample article from our corpus (including 7 Canadian and 2 Australian location expressions). Annotation includes string span, notation, latitude, and longitude. The last Melbourne in the HTML text of this article had a hyperlink to another Wikipedia article, resulting in different coordinates annotated. Note that location expressions without hyperlinks (e.g., Australia in the sample) are ignored in the annotation.
• Figure 3: Coordinates that appear more than 10 times in the WHLL corpus.
• Figure 4: Dependency-based strategy.
• Figure 5: Geocoding accuracy to tolerance error distance.