Finding Bent-double Radio Galaxies: A Case Study in Data Mining

Abstract

This paper presents our early results in applying data mining techniques to the problem of finding radio-emitting galaxies with a bent-double morphology. In the past, astronomers on the FIRST (Faint Images of the Radio Sky at Twenty-cm) survey have detected such galaxies by first inspecting the radio images visually to identify probable bent-doubles, and then conducting observations to confirm that the galaxy is indeed a bent-double. Our goal is to replace this visual inspection by a semi-automated approach. In this paper, we present a brief overview of data mining, describe the features we use to discriminate bent-doubles from non-bent-doubles, and discuss the challenges faced in defining meaningful features in a robust manner. Our experiments show that data mining, using decision trees, can indeed be a viable alternative to the visual identification of bent-double galaxies.