Inspecting Influences on Likes and Comments of Photos in Instagram

摘要

Instagram receives millions of photos every day. While many photos get a lot of likes and comments, others do not get much attention. This paper analyzes what factors lead to these differences, and what photos with different levels of popularity look like. To examine the influences to the number of likes and comments, we introduce the publish time, user social context and post content. We then make statistical analysis based on single factor, as well as machine learning based analysis with several combinations of factors. By qualitative and quantitative analysis, we gain some favorable conclusions. The time factors at different scale have different effects. The number of likes and comments vary from year to year while they look similar on each day of a week. Moreover, posts published after noon tend to receive more attention. The number of followers of users plays the most important role in user social context, while the semantic feature of caption and the visual feature of photos are the main factors of post content affecting the number of likes and comments.