We present the results of applying automated machine learning techniques to the problem of matching different object catalogues in astrophysics. In this study we take two partially matched catalogues where one of the two catalogues has a large positional uncertainty. The two catalogues we used here were taken from the HI Parkes All Sky Survey (HIPASS), and SuperCOSMOS optical survey. Previous work had matched 44% (1887 objects) of HIPASS to the SuperCOSMOS catalogue. A supervised learning algorithm was then applied to construct a model of the matched portion of our catalogue. Validation of the model shows that we achieved a good classification performance (99.12% correct). Applying this model, to the unmatched portion of the catalogue found 1209 new matches. This increases the catalogue size from 1887 matched objects to 3096. The combination of these procedures yields a catalogue that is 72% matched.