The massive parked domains exist in the Internet, which seriously affect the Internet experience and Internet environment of online users when surfing. In order to recognize parked domains, a new method of parked domain recognition was proposed based on authoritative Domain Name Server (DNS). The set of authoritative DNS which could be used for domain parking service was extracted by the typosquatting domains commonly used in domain parking service. Then the set was clustered by semi-supervised clustering method to identify the authoritative DNS associated with domain parking service. When detecting a parked domain, the parked domain was recognized by the judgments that whether its authoritative DNS was applied in domain parking service and whether its mapped IP addresses was concluded in the set of IP addresses of parking Web servers. By using the existing detecting method based on webpages' features, the accuracy of the proposed method was analyzed. The experimental results show the proposed method has achieved the accuracy rate of 92.8%, and avoids crawling the webpage information, which has a good performance on parked domains detection in real-time.