Automatic Extraction of Destinations, Origins and Route Parts from Human Generated Route Directions

  • Xiao Zhang
  • Prasenjit Mitra
  • Alexander Klippel
  • Alan MacEachren
Conference paper

DOI: 10.1007/978-3-642-15300-6_20

Part of the Lecture Notes in Computer Science book series (LNCS, volume 6292)
Cite this paper as:
Zhang X., Mitra P., Klippel A., MacEachren A. (2010) Automatic Extraction of Destinations, Origins and Route Parts from Human Generated Route Directions. In: Fabrikant S.I., Reichenbacher T., van Kreveld M., Schlieder C. (eds) Geographic Information Science. GIScience 2010. Lecture Notes in Computer Science, vol 6292. Springer, Berlin, Heidelberg

Abstract

Researchers from the cognitive and spatial sciences are studying text descriptions of movement patterns in order to examine how humans communicate and understand spatial information. In particular, route directions offer a rich source of information on how cognitive systems conceptualize movement patterns by segmenting them into meaningful parts. Route directions are composed using a plethora of cognitive spatial organization principles: changing levels of granularity, hierarchical organization, incorporation of cognitively and perceptually salient elements, and so forth. Identifying such information in text documents automatically is crucial for enabling machine-understanding of human spatial language. The benefits are: a) creating opportunities for large-scale studies of human linguistic behavior; b) extracting and georeferencing salient entities (landmarks) that are used by human route direction providers; c) developing methods to translate route directions to sketches and maps; and d) enabling queries on large corpora of crawled/analyzed movement data. In this paper, we introduce our approach and implementations that bring us closer to the goal of automatically processing linguistic route directions. We report on research directed at one part of the larger problem, that is, extracting the three most critical parts of route directions and movement patterns in general: origin, destination, and route parts. We use machine-learning based algorithms to extract these parts of routes, including, for example, destination names and types. We prove the effectiveness of our approach in several experiments using hand-tagged corpora.

Keywords

driving directions route component classification destination name identification geographic information extraction 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer-Verlag Berlin Heidelberg 2010

Authors and Affiliations

  • Xiao Zhang
    • 1
  • Prasenjit Mitra
    • 1
    • 2
  • Alexander Klippel
    • 3
  • Alan MacEachren
    • 3
  1. 1.Department of Computer Science and Engineering 
  2. 2.College of Information Science and Technology 
  3. 3.Department of GeographyThe Pennsylvania State University 

Personalised recommendations