Abstract: The exponential growth of social network users makes it difficult for users to find suitable friends in the network. Therefore, this paper proposes a microblog friend recommendation algorithm based on SSD and timing model. Firstly, the multi-target detection algorithm SSD is used to extract the information of the collected users’ images. Secondly, the timing model is used to further process the extracted information in time dimension. Then, JS divergence formula is used to calculate the similarity between users. Finally, the weighted fusion is carried out with the similarity based on the user’s personal information to obtain the comprehensive user similarity, and the Top-K idea is used for user re- commendation. The verification on Sina microblog users’ data set shows that the weight value of refe- rence factors will affect the recommendation results and the precision of this method is higher than that of the algorithm only considering users’ attributes or users’ pictures.

Key words: social network, target detection, friend recommendation, timing model