Member of the International Virtual Observatory Alliance

RASS Source Identification with ClassX

The ClassX team has developed a set of tools for the classification of ROSAT sources (McGlynn, Suchkov et al., 2004). The tools are oblique decision tree classifiers using the OC1 implementation. The classifiers come in various flavours, depending on the amount of information about the source that must be provided, which is directly related to the information used in teaching the classifiers. Most of the classifiers require more information than can be obtained from the ROSAT data itself. For example some require information from optical, infrared and/or radio wavebands. To obtain this information therefore requires an identification between the ROSAT sources and sources in other catalogues.

In the GAVO prototype discussed here we use one particular classifier that has been made available via HTTP POST by the ClassX team. This classifier requires an identification between the ROSAT sources and the USNO-B1.0 catalogue for optical information. It further requires an indicator of whether an identification can be made with a source in either the SUMSS or the NVSS radio catalogues. All these catalogues are published by VizieR, which offers a "list upload service" which is very useful for our purposes. This service allows one to upload a list of positions and it will do a cone search around each of the positions, and the application makes extensive use of this feature.

NOTE: This service is no longer available.


We thank the ClassX team for their cooperation in this project, We especially thank Erik Winter for frequent and quick help with the ClassX web service.


  • T.A. McGlynn, A.A. Suchkov, E.L. Winter, R.J. Hanisch, R.L. White,
    F. Ochsenbein, S. Derriere, W. Voges, M.F. Corcoran, S.A. Drake, and M. Donahue
    Automated Classification of ROSAT Sources Using Heterogeneous Multiwavelength Source Catalogs
    submitted for publication in Astrophysical Journal, 2004