Astronomical Data Mining

Abstract

Various research projects have been conducted in an attempt to explore and improve the classification process of astronomical images as well as enhance the study of the classified data. The growing interest in this area is, to a large extent, attributed to the introduction of citizen science projects like Galaxy Zoo that host copious amounts of such data and encourage the public to involve themselves in classifying and categorising these images. It is also due, in large part, to the profusion of data being collated by numerous sky surveys like the Sloan Digital Sky Survey (SDSS) which, at present, hosts an imaging catalogue of over 350 million objects and is continuously growing [4].With such copious amounts of data, it would not be unreasonable to state that our capacity today for acquiring data has certainly far outstripped our capacity to analyse it, there is no doubt that manual processing has long become impractical, creating a need for automated methods of analysis and study. This is where data mining comes in, thus creating a new paradigmatic approach, dubbed fairly recently as the fourth paradigm [19, 77]. Data mining has emerged as an important field of study at the time of convergence of large data repositories’ sizes and the computational power of the high performing computational facilities [62]. Finding its roots in machine learning, pattern recognition, statistics, and databases, data mining has grown tremendously over the past two decades. Recently, data mining faces the challenge of the ever increasing size of data repositories, thanks to advances in both hardware and software technologies. Among the very large data repositories come the astronomical data produced by sky surveys.