In this study, we employ machine learning to build a catalog of DB white dwarfs (DBWDs) from the LAMOST Data Release (DR) 5. Using known DBs from SDSS DR14, we selected samples of high-quality DB spectra from the LAMOST database and applied them to train the machine learning process. Following the recognition procedure, we chose 351 DB spectra of 287 objects, 53 of which were new identifications. We then utilized all the DBWD spectra from both SDSS DR14 and LAMOST DR5 to construct DB templates for LAMOST 1D pipeline reductions. Finally, by applying DB parameter models provided by D. Koester and the distance from Gaia DR2, we calculated the effective temperatures, surface gravities and distributions of the 3D locations and velocities of all DBWDs.