Aiming at the actual needs of content recommendation of the Army Equipment Information Network, combined with the characteristics of users and content on the website, this article analyzed that the traditional collaborative filtering algorithm has poor recommendation results due to cold start and data sparse. A new algorithm which combined content-based recommendation algorithm with traditional collaborative filtering algorithm was proposed to solve this problem. The authors researched and designed a suitable model for the content recommendation of the website, and proposed a proper recommendation algorithm. Finally, the content of the website and the users’ behavior data were used as experimental content to verify the validity of the proposed model algorithm. The results showed that the proposed recommendation algorithm could maintain a certain accuracy and fraction of coverage when the K value was at an appropriate value. The method met the personalized needs of users when browsing the website news.