Object Positioning Based on Partial Preferences
- 634 Downloads
In several situations, a set of objects must be positioned based on the preferences of a set of individuals. Sometimes, each individual can/does only include a limited subset of objects in his preferences (partial preferences). We present an approach whereby a matrix of distances between objects can be derived based on the partial preferences expressed by individuals on those objects. In this way, the similarities and differences between the various objects can subsequently be analyzed. A graphical representation of objects can also be obtained from the distance matrix using classical multivariate techniques such as hierarchical classification and multidimensional scaling.
KeywordsPreference structures Object representation Multivariate analysis Classification
Unable to display preview. Download preview PDF.
- 2.Busing, F., Commandeur, J., Heiser, W.J.: PROXSCAL: A multidimensional scaling program for individual differences scaling with constraints. In: Bandilla, W., Faulbaum, F. (eds.) Softstat 1997: Advances in Statistical Software, vol. 6, pp. 67–74. Lucius & Lucius, Stuttgart (1997)Google Scholar
- 3.Commandeur, J., Heiser, W.J.: Mathematical derivation in the proximity scaling (PROXSCAL) of symmetric data matrices. Research Report RR-93-04, Department of Data Theory, Leiden University (1993)Google Scholar
- 5.Everitt, B.S.: Cluster Analysis. Edward Arnold, London (1993)Google Scholar
- 8.Legendre, P., Legendre, L.: Numerical Ecology, 2nd English edn. Elsevier, Amsterdam (1998)Google Scholar
- 10.Spearman, C.: Footrule for measuring correlation. British Journal of Psychology 2, 89–108 (1906)Google Scholar
- 11.Van de Walle, B.: A Relational Analysis of Decision Makers’ Preferences. Int. J. of Intel. Systems 18, 175–791 (2003)Google Scholar
- 13.Wu, Z., Meng, W., Yu, C., Liz, Z.: Towards a higly scalable and effective metasearch engine. In: Proc. 10th Int. Web Conf., Hong-Kong, pp. 386–395 (2001)Google Scholar