Probability issues in locality descriptions based on Voronoi neighbor relationship

  • Authors:

  • Yongxi Gong;Lun Wu;Yaoyu Lin;Yu Liu

  • Affiliations:

  • Institute of Remote Sensing & Geographical Information Systems, Peking University, Beijing 100871, PR China and School of Urban Planning and Management, Harbin Institute of Technology Shenzhen Gra ...;Institute of Remote Sensing & Geographical Information Systems, Peking University, Beijing 100871, PR China;School of Urban Planning and Management, Harbin Institute of Technology Shenzhen Graduate School, Shenzhen 518055, PR China and State Key Laboratory of Subtropical Building Science, South China Un ...;Institute of Remote Sensing & Geographical Information Systems, Peking University, Beijing 100871, PR China

  • Venue:
  • Journal of Visual Languages and Computing
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Spatial relationships play an important role in spatial knowledge representation, such as in describing localities. However, little attention has been paid to how to describe the position of a target object (TO) with a qualitative referencing system that consists of a set of reference objects (ROs) in the locality description context. We propose a method that accounts for the differences between two scenarios in locality descriptions. This method is probabilistic and is based on the Voronoi neighbor relationship to determine candidate ROs for describing a given TO's position in the second scenario. The Voronoi neighbor relationship is adopted to determine candidate ROs of a TO and to compute the neighboring area of an RO. A probability function is presented to model the uncertainty of selecting appropriate ROs. To build locality descriptions that are consistent with commonsense, four constraints are placed on the probability function. Two probability functions based on Euclidean distance and stolen-area, and a mixed probability function that considers both Euclidean distance and stolen-area, are analyzed and compared. With the mixed probability function, we establish a method to construct the locality description of a given TO. Finally, three examples demonstrate how to select ROs to describe a TO's position.