Tutorial 2: Behavioral Similarity Measures

Behavioral Similarity Measures for Spatial Cognition

 by  Alexander Klippel, Penn State University, PA, USA

Abstract

Similarity measures are central to human cognition (Tversky, 1977) and have long been a central aspect for modeling spatial knowledge (Tobler, 1970). To determine the similarity of objects and events from a cognitive perspective several approaches have been proposed, specifically: geometric, feature-based, alignment based, and transformational (Goldstone, 1999). Among these measures, the geometric approach has gained widespread attention (Gärdenfors, 2000; Ahlqvist, 2004; Schwering & Raubal, 2005; McIntosh & Yuan, 2005). In all disciplines participating in the endeavor of cognitive science, geometric approaches are used to model similarity and conceptual knowledge, by applying distance concepts such as network distance or Euclidean distance (Widdows, 2004). Similarity is inversely proportional to distance: the smaller the distance between two entities, the more similar they are. Similarity has gained widespread interest in spatial sciences as a necessity for formally characterizing categorization. However, behavioral studies on the similarity of spatio-temporal events have not gained the same attention as formal treatments. Partially, this is due to the difficulties with perceived similarity that have long been an argument in psychology (for an overview see Goldstone, 1994). The tutorial will provide coverage of the topic addressing three key issued: 1) A (necessarily) brief introduction to similarity and similarity measures; 2) a detailed discussion of behavioral approaches to assessing similarity; 3) the analysis of similarity measures as a result from experiments with a special focus on cluster analysis. Part 2 and 3 will be accompanied by demonstrations using a purpose built software tool (Klippel, Worboys & Duckham, to appear) that allows for the grouping of animated icons and available software solutions for analyzing similarity measures (CLUSTAN and SPSS). The goal of the tutorial is to provide researchers interested in spatial cognition with a deeper understanding of similarity measures, which behavioral methods are available to assess similarity, which tools exist to conduct behavioral research on similarity, and which analysis techniques can be used.

Contents

The workshop will offer theoretical and hands-on knowledge on the following topics:

• A (necessarily) short introduction to similarity measures from a cognitive perspective
  
• The importance of similarity for cognitive processes
• Questions of perceptual and conceptual similarity
• Models describing similarity
• Similarity measures for spatio-temporal information
  
• Design aspects of experiments eliciting similarity measures
• Classic card sorting
• Card sorting using Euclidean distance
• Direct similarity ratings
• Similarity validation
• Similarity selection
  
• Hands-on knowledge on analyzing similarity measures obtained from behavioral experiments
• Some basics of clustering analysis
• Different hierarchical clustering methods
• Using SPSS and CLUSTAN to analyze similarity matrices
• Cluster validation
  

Audience

Behavioral similarity measures in spatial cognition are interesting to a number of discipline, and analysis techniques such as cluster analysis can be found in several recent publications (Makany, Redhead & Dror, 2007; Hölscher, Meilinger, Vrachliotis, Brösamle & Knauff, 2005; Xu, 2007; Nedas, Egenhofer & Wilmsen, 2007; Schwering 2008). The workshop is designed for researches interested in conceptual knowledge elicitation, for example researchers interested in ontologies; researchers interested in identifying similarities across different spatial behaviors in psychology and geographic information science; researchers interested in evaluating formal characterizations of spatial knowledge through behavioral studies.

Instructor

Dr. Klippel received his Ph.D. from the SFB/TR 8: Spatial Cognition in 2003. He has worked on several aspects of spatial cognition at the interface of behavioral research and formal characterization of spatial knowledge. Dr. Klippel is teaching courses on spatial analysis and GeoSemantics. The GeoSemantics course has first been developed in Melbourne, Australia, and is now being held at Penn State. Dr. Klippel is Directing the Human Factors in Geographic Information Science Laboratory (HFgisL) at the GeoVISTA Center. A strong focus of this lab is placed on the development and evaluation of interfaces and cognitive factors in GIScience. Research at the qualitative / quantitative interface, especially research focusing on spatio-temporal aspects, is central to the development of, and building of theory within geographic information science. Behavioral methods using similarity are currently employed to evaluate characteristics of multivariate visualization techniques.

References

UNDER CONSTRUCTION