The Amazon website generates recommendation lists for each book, allowing users to select the books they are interested in. But this recommendation method only considers the similarity among books. Based on the recommendation lists, this paper introduces a similarity of book recommendation lists, and calculates Jaccard similarity of different book recommendation lists, then generates recommendation list for each book by similarity ranking. This paper combines it and the recommendation list provided by Amazon to conduct personalized recommendation for users. The experimental results show that there is a certain improvement in the average accuracy rate, the average recall rate, Macro_F1 and Micro_F1 compared to using only the recommendation list provided by Amazon. It can be seen that the similarity of the recommendation list of the book can play a certain role in optimizing the recommendation effect.