The invention discloses a burst feature detection method and apparatus used for microblog, relates to the technical field of network information mining, and aims to solve the problems of large microblog information noise and low data diversity burst feature detection accuracy in the prior art. The method comprises the steps of detecting candidate burst features according to an instant burst situation of a detected feature of a meaningful string in an information stream; and filtering frequent dummy burst features and intermittent dummy burst features in the candidate burst features.