摘要

For decision tables with a very small number of objects, this paper introduces a model combining the equivalence relations of rough set theory and algebraic structures. As an example application, we classify six-dimensional attribute vectors (maintenance quality indices) in a sample of 55 bridges from Chongqing province in China. A panel of experts has already used these data to rank the sites in terms of overall management quality, and full use is made of their decisions in training the model. Compared with the performances of a classical rough set model and a support vector machine, the new model is shown to be both feasible and more accurate.