Postdoc in Question-based Analysis of Geographic Information with Semantic Queries (0.8 FTE)

In the age of big data, geographic information has become a central means for data scientists of various disciplines to embed their analysis into a spatio-temporal context, from human mobility patterns and social inequality to the investigation of personal health. However, as the variety of data sources and software available on the Web increases, it becomes more and more impossible to comprehend and utilize all tools and data available to answer geo-analytical questions. Hence, whenever a functionality is needed but not available in one tool, analysts are forced to reformulate their questions in terms of technicalities of another tool or other datasets. This procedure does not scale with the increasing variety of geo-analytic sources on the web, preventing analysts from tapping its full potential. Consider, in contrast, how easy it is for a user of a digital smartphone assistant such as Amazon's Alexa to ask a question like "What is the weather today?" and get back an answer from the Web. It would mean a tremendous breakthrough if analysts could similarly ask questions in order to get the tools and data required to answer them. Unfortunately, geo-analytic technology currently cannot handle questions.

To realize this vision, it is necessary to understand how geo-analytic resources can be captured with the questions they answer. The QuAnGIS project, a 5-year research project at Utrecht University, starting in January 2019 funded by the European Research Commission (ERC), develops a theory about interrogative spatial concepts needed to turn geo-analytical questions into a machine-readable form using Semantic Web and Question-Answering (QA) technology. Concepts include "spatial core concepts" field, object, network, event (Kuhn 2012), as well as analytical concepts such as accessibility, exposure, density, distance and aggregation. In this form, questions can be matched with the capacity of major analytical GIS tools and data sources on the Web.

We are looking for an experienced Researcher in order to cope with the conceptual as well as computational challenges involved and to help guide the project team (2 PhD candidates and a technician) together with the project leader. The Postdoc is responsible for developing technology to match spatial questions in the form of semantic queries with descriptions of geospatial tools and data sources. This is the central computational mechanism of the developed technology stack.

The Postdoc will work on the following topics:
? help develop a gold standard of question-answers in geospatial analysis for evaluation;
? develop tractable query language (SPARQL) subsets for query matching;
? develop matching algorithms and reasoning technology for matching questions and tools;
? evaluate question matching on gold standard;
? help perform user study for evaluating technology stack.

Tasks of the Postdoc consist of (but are not limited to):
? conducting and spurring scientific research in the fields of Geospatial Semantics, Geographic Information Science, Ontology Engineering, or Semantic Web relevant to handling spatial questions and analytic resources;
? collection of data for developing a gold standard of question-answers as well as analytic (web) resources for geospatial analysis (filling web repository with content);
? perform user studies for specifying test scenarios and for testing the technology stack, including the matching algorithm, the question formulation interface as well as the tool/data repository with geospatial analysts;
? publication of results in scientific journals and presentation in high quality international conferences in these fields;
? organizing multi-stakeholder meetings and workshops to test and discuss the technology;
? to develop teaching skills, researchers in the project are expected to contribute to the teaching programme of the Department of Human Geography and Spatial Planning to a limited degree (up to 10%).