Font Size:  Small  Medium  Large

Deconstruction of geo-analytical questions in terms of measures, supports, and spatio-temporal extents

Enkhbold Nyamsuren, Haiqi Xu, Simon Scheider, Eric J Top, Niels Steenbergen


This study investigates the GeoAnQu corpus of geo-analytical questions. Unlike other question corpora, the questions in this corpus imply analytical goals and are thus supposed to be answered with GIS workflows, not with the retrieval of geographic facts. We investigate how geo-analytical questions are structured syntactically and semantically, and how the structure may be interpreted by human analysts to compose workflows. Our question analysis model is based on the notions of a measure, support, and extent, which are inspired by Sinton's three dimensions of spatial analysis. We use XPath queries to automatically extract syntactic patterns from constituency parse trees corresponding to these notions. Results show that geo-analytical questions are of considerable complexity, yet often have predictable syntactic patterns that can be reliably mapped to measures, supports, and extents. Furthermore, we identify analytical goals attributable to these notions. To our knowledge, this is the first reported systematic analysis of this kind. The findings open new opportunities in Natural Language Interpretation and query generation for the automated answering of geo-analytical questions. Additionally, our study shows that questions asked in a scientific context can be on different levels of concreteness. Therefore, we also discuss best practices for formulating questions clearly and concretely.

Full Text: PDF

Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 License.