Histogram ranking with generalised similarity-based TOPSIS applied to patent ranking

摘要

In this paper, we introduce a new version of the well-known TOPSIS method: similarity-based TOPSIS. The new method uses a similarity measure to replace the commonly used distance measure. Similarity measure is formed under generalised Lukasiewicz structure that allows us to form the measure in a more general structure and this way enhances (patent) ranking results. At the same time, however, the selection of a similarity parameter becomes a problem. To fix this new problem, a new method that we call histogram ranking is introduced. Histogram ranking is usable for relaxing the dependence of ranking on parameter value; it is designed to be a complement to parameter dependent ranking methods and is usable, when it is difficult to select precise parameter values. Histogram ranking is based on calculating the centre of gravity points from the histograms and this information is then used to form parameter value independent ranking of the object.