Preselecting AGN candidates from multi-wavelength data by ADTree

With the information era in astronomy coming, this "data avalanche" may provide many answers to important problems in contemporary astrophysics. The most important problem is sifting through massive amounts of data to mine knowledge. In this paper, we positionally cross-identify multi-wavelength data from optical, near-infrared, and x-ray bands, and then employ alternating decision trees (adtree) to quickly and robustly separate AGN candidates to a high degree of accuracy. We emphasise the application of the method due to the development of large survey projects and the establishment of the virtual observatory, and conclude that the application of data mining algorithms in astronomy is of great importance to discover new knowledge impossible to obtain before, and promote the development of astronomy.