Application of Computer Algorithm in New Media Content Recommendation System

Nowadays, the information acquisition channels of media have gradually shifted to the Internet, and the decline of traditional media has become a foregone conclusion in the face of new media backed by big data, and the strong rise of new media has also changed the mode of information push. This personalized push mechanism has always had the problem of poor alignment with the customer's interest points. Therefore, this paper proposes to optimize the content recommendation system by using the computer algorithm represented by the Bayesian classification algorithm, which not only improves its performance, but also tries to alleviate the information cocoon phenomenon brought by the current new media. In the final comparison experiment on the average absolute error between the user's predicted rating and the actual rating, the average absolute error of the experimental group based on the proposed method was 0.2974, while the average absolute error of the control group using the conventional system was 0.3481. In other words, this method greatly improves the control of the content recommendation system over the user's preferences, so as to push the content that they are more interested in.