ИСТИНА

Traditionally, data mining in large surveys implies sophisticated statistical analysis of large volumes of scientific measurements. In astronomy, this usually yields statistically significant or complete samples of objects to study their properties or looking for something new and rare. I will present the extension of this paradigm: (i) we first re-analyze large volumes of data, specifically, millions of spectra of galaxies and quasars originating from modern wide-field surveys (SDSS, SDSS-V, LAMOST, DESI) using a novel approach and publish these results as a publicly available Virtual Observatory resource; (ii) then exploit classical data mining techniques to identify samples of rare objects; (iii) then follow-up the most promising sources using world's most advanced observational facilities on the ground (Keck, VLT, MeerKAT) and in space (Chandra, XMM-Newton, HST, JWST) to unveil their nature and provide astrophysical interpretation. I will present a few prominent examples, which include the discovery of a population of intermediate-mass black holes, establishing the nature of ultra-diffuse galaxies in clusters, and our recent result on the identification of low-redshift analogs of enigmatic `Little Red Dots' discovered in the early Universe by JWST surveys.