A Dataset and Evaluation Framework for Complex Geographical Description Parsing

Proceedings of the 28th International Conference on Computational Linguistics, pages 936–948 Barcelona, Spain (Online), December 8-13, 2020 936 A Dataset and Evaluation Framework for Complex Geographical Description Parsing Egoitz Laparra School of Information University of Arizona Tucson, AZ, USA laparra@email.arizona.edu Steven Bethard School of Information University of Arizona Tucson, AZ, USA bethard@email.arizona.edu Abstract Much previous work on geoparsing has focused on identifying and resolving individual toponyms in text like Adrano, S.Maria di Licodia or Catania. However, geographical locations occur not only as individual toponyms, but also as compositions of reference geolocations joined and modified by connectives, e.g., “...between the towns of Adrano and S.Maria di Licodia, 32 kilometres northwest of Catania”. Ideally, a geoparser should be able to take such text, and the geographical shapes of the toponyms referenced within it, and parse these into a geographical shape, formed by a set of coordinates, that represents the location described. But creating a dataset for this complex geoparsing task is difficult and, if done manually, would require a huge amount of effort to annotate the geographical shapes of not only the geolocation described but also the reference toponyms. We present an approach that automates most of the process by combining Wikipedia and OpenStreetMap. As a result, we have gathered a collection of 360,187 uncurated complex geolocation descriptions, from which we have manually curated 1,000 examples intended to be used as a test set. To accompany the data, we define a new geoparsing evaluation framework along with a scoring methodology and a set of baselines. 1 Introduction Geoparsing, or toponym resolution, is the task of identifying geographical entities, and attaching them to their corresponding reference in a coordinate system (Gritta et al., 2018b). In its traditional setting it Figure 1: An illustrative example of complex geographical description parsing. Given a text describing a target geolocation the goal is to aproximate its geometry using as reference the geometries of the geolocations that appear in the description (Adrano, S.Maria di Licodia and Catania). In this example, the target geolocation is Biancavilla, but in general, there may not be a name for the target geolocation. This work is licensed under a Creative Commons Attribution 4.0 International License. License details: http:// creativecommons.org/licenses/by/4.0/. 937 involves two main steps: parsing the text to recover geo-location mentions and linking those mentions to the entries of a toponym dictionary like GeoNames1. Thus, it can be seen as a special case of named entity recognition and disambiguation. The task plays a key role in information extraction, since it allows events to be placed in an unequivocal location that could be plotted on a map. However, the setting described above restricts the task to geolocations with explicit names, whilst text in natural language can refer to geographical regions by complex descriptions that involve a set of different toponyms. Take, for example, the following snippet: “a town and comune in the Metropolitan City of Catania, Sicily, southern Italy. . . located between the towns of Adrano and S.Maria di Licodia, 32 kilometres (20 …

Content truncated to 500 words