Abstract: As a powerful means of information, big data technology and artificial intelligence technology are moving toward integration. They provide brand-new information resources, technical means and governance paradigms for the network public opinion management. And to a certain extent, they have realized the comprehensive, automatic, scientific, precise and personalized management of network public opinion. However, there still exist some challenges like deviations between the public opinion presented by the algorithm and the real public opinion, public opinion misjudging risks, false news, information cocoon rooms, public opinion misleading in decision-making, privacy ethics, data monopoly, and risks of data abuse in information management. In order to solve the challenges, some measures for improvement are concluded in the paper.