Artificial intelligence is one of the many branches of computer science. With the continuous development of information technology and computer networks, artificial intelligence technology has also received tremendous support and rapid development. In the computer field, recommendation systems can be said to be the most extensive and successful application of artificial intelligence technology. This paper focuses on the problems of cold start and low recommendation accuracy caused by sparse rating data in traditional recommendation systems, proposes a personalized recommendation algorithm based on online comment sentiment analysis. The algorithm uses feature-level sentiment analysis to mine user preference information implicit in comments, and finally implements recommendations based on user ratings and preference information. Experiments show that the algorithm proposed in this paper can effectively make up for the shortcomings of traditional collaborative filtering recommendation algorithms, and is of great help in improving user cold start and the quality of recommendation results.