Author(s)

S. Dzeroski, L. Dehaspe, B.M. Ruck & W.J. Walle

Abstract

We use machine learning to find rules for the direct interpretation of bio- logical samples of river life. Given are approximately three hundred field samples of bent hie communities taken from the upper Trent catchment in England, and classified into water quality terms by an expert river ecologist (H. A. Hawkes). From these, we induce classification rules that capture the expertise of the ecologist. Different machine learning techniques are em- ployed to this end, including rule induction in prepositional and first-order logic. The rules were inspected by the ecologist and found to be mostly consistent with his knowledge. Biological classification of water quality Biological monitoring of river water quality is of increasing im

Keywords