Project

Back to overview

Improving Access to Unstructured Text through "Folk-centred" Approaches (FolkOnt2)

Applicant Purves Ross
Number 146157
Funding scheme Project funding (Div. I-III)
Research institution Geographisches Institut Universität Zürich
Institution of higher education University of Zurich - ZH
Main discipline Other disciplines of Environmental Sciences
Start/End 01.04.2013 - 30.09.2013
Approved amount 29'860.00
Show all

All Disciplines (2)

Discipline
Other disciplines of Environmental Sciences
Other disciplines of Engineering Sciences

Keywords (8)

Geographic Information Science; Geographic Information Retrieval; folksonomy; semantics; vernacular placenames; ambiguity; vagueness; ontology

Lay Summary (German)

Lead
Wie beschreiben Menschen ihre Umwelt? Es gibt theoretische Überlegungen, die auf psychologischen Untersuchungen und philosophischen und linguistischen Lehren basieren und nahelegen, dass der Mensch die natürliche Umwelt in konkrete Objekte unterteilt, und dass diese Objekte in der Kommunikation eine zentrale Rolle spielen. Beispiele von solchen Objekten sind Berge, Täler oder Seen.
Lay summary

In der Theorie wird betont, dass die Art und Weise wie die Umwelt in Objekte gegliedert wird von Individuum zu Individuum, und zwischen sozialen Gruppen und Kulturen stark variieren kann. Darum können lokale Klassifikationen stark von offiziellen, von Experten geprägten Taxonomien, abweichen.

Lokale Variation bedeutet eine grosse Herausforderung für die Vereinheitlichung von räumlichen Daten einerseits, und andererseits für die Kommunikation zwischen Experten und Laien. Es gibt zwar eine Reihe von ethnographischen Untersuchungen die gezielt lokale Umweltbeschreibungen dokumentieren, solche Untersuchungen haben aber oft nur limitierte räumliche und zeitliche Ausdehnung.

Unser übergeordnetes Ziel ist es lokale Unterschiede in Kategorisierungen der Umwelt aus Textbeschreibungen automatisch zu extrahieren und damit grosse räumliche Ausschnitte und lange Zeitspannen abdecken zu können. Dazu müssen wir:

 (i) Umweltbeschreibungen automatisch räumlich referenzieren können, und

 (ii) diese referenzierten Beschreibungen auf relevante Information untersuchen und diese explizit erfassen. 

Es ist wichtig, dass wir beim extrahieren von Information berücksichtigen, dass grosse Variationen und sogar Widersprüche in der Umweltbeschreibung erhalten bleiben.

Unsere Arbeit wird neue Methoden generieren, mit denen lokale Umweltbeschreibungen auf grossen zeitlichen und räumlichen Skalen erfasst und verglichen werden können. Wir erwarten, dass solche Information zentral ist um zwischen Experten und Laien zu vermitteln und, um Suchmaschinen um eine räumliche Komponente zu erweitern, die eben auch räumliche Spezialitäten berücksichtigt.

Direct link to Lay Summary Last update: 06.11.2013

Responsible applicant and co-applicants

Employees

Publications

Publication
From Space to Place: Place-Based Explorations of Text
(2015), From Space to Place: Place-Based Explorations of Text, in International Journal of Humanities and Arts Computing, 9(1), 74-94.
Creating Test Collections from User Generated Content for GIR Evaluation
Palacio Damien, Derungs Curdin, Purves Ross, Creating Test Collections from User Generated Content for GIR Evaluation, in Proceedings of the 7th Workshop on Geographic Information Retrieval, Orlando, Florida.

Collaboration

Group / person Country
Types of collaboration
Chris Jones, Cardiff University Great Britain and Northern Ireland (Europe)
- in-depth/constructive exchanges on approaches, methods or results
- Publication
David Mark, State University of New York, Buffalo United States of America (North America)
- in-depth/constructive exchanges on approaches, methods or results

Scientific events

Active participation

Title Type of contribution Title of article or contribution Date Place Persons involved
7th Workshop on Geographic Information Retrieval Talk given at a conference Creating Test Collections from User Generated Content for GIR Evaluation 05.11.2013 Orlando, Florida, United States of America Derungs Curdin; Weibel Robert; Purves Ross;
Digital Texts and Geographical Technologies in the Digital Humanities Individual talk From space to place: place-based explorations of text 08.07.2013 Lancaster, UK, Great Britain and Northern Ireland Purves Ross;
Landscape through the prism of language Individual talk Delineating and describing landforms in geographic information science 01.06.2013 Zurich, Switzerland Purves Ross;


Associated projects

Number Title Start Funding scheme
126659 Improving Access to Environmental Databases through "Folk-centred" Ontologies (FolkOnt) 01.04.2010 Project funding (Div. I-III)
149397 Computational methods for extracting landmark semantics 01.09.2013 International short research visits

Abstract

In this proposal we request a modest 6 month extension of SNF project FolkOnt (No. 200021-126659), a three year project which commenced on 01.04.2010. This extension will be used to consolidate existing results and take advantage of a new dataset which has emerged as an excellent testbed for our methods during the first 2.5 years of the project.The FolkOnt and FolkOnt II project are based in the domains of Geographic Information Re-trieval (GIR) and, in particular the use of methods from GIR to explore landscape characterisation. Thus, our research also contributes to understandings of how landscape descriptions vary in both space and time, and at a broader level seeks to develop methods which improve access to unstructured data for both expert and lay users.The overall aim of FolkOnt and FolkOnt II can be stated as follows:"How can the vagueness and ambiguity present in unstructured descriptions of natural landscapes, and particularly protected areas, be captured such that geographic and conceptual queries are effectively resolved for both expert and lay communities?"To date, the project has developed methods which allow us to distinguish between and disambiguate fine granularity toponyms referring to natural features, and link corpora of natural language descriptions to space. We have therefore been able to explore how natural landscape descriptions have varied in space and time in a Swiss corpus. Work has been presented at number of conferences including the American Association of Geographers (2011), Geocomputation (2011), Swiss GeoScience Meeting (2011), AGILE 2012, Congress of the International Society for Dialectology and Geolinguistics (2012) and GIScience 2012. Furthermore, a journal paper, based on a related Masters thesis has been published, and another is currently in revision. Work in the project extension will focus on two issues. Firstly, we will develop methods to explicitly link the folksonomic representation of landscape terms thus far extracted from our text corpus with geographic data describing landscape variation in Switzerland (for example Corine and Arealstatistik). Secondly, we will complete development of a system showcasing both how the methods in FolkOnt can be used to retrieve information spatially and thematically, and how such information can be used to characterise geographic regions. The project will be hosted in the Geocomputation Group at the Department of Geography at the University of Zurich, but will continue to take the form of collaborative research with the Data Centre Nature and Landscape (DNL) of Swiss Federal Institute for Forest, Snow and Landscape Research.
-