Big data will inevitably have a profound impact on public security intelligence science and its research paradigm. Public security intelligence science needs to make use of big data to improve its research paradigm, and to avoid negative effects of big data on its development. Therefore, under big data environment, the existing research paradigm of public security intelligence science (intelligence-led policing) must be adjusted: data being one of the research objects, intelligence application becoming important research content, and the multidisciplinary research methods being introduced into the science. The research paradigm, with big data integrated, should avoid the problems of the latter: relying on correlation analysis but ignoring causality analysis; advocating universal data but despising people's intelligence; worshiping technology but neglecting the dominant position of human being.