"Woman-Metal-White vs Man-Dress-Shorts": Combining Social, Temporal and Image Signals to Understand Popularity of Pinterest Fashion Boards

Abstract: Pinterest is a popular photo sharing website. Fashion is one the most popular and content generating category on this platform. Most of the popular fashion brands and designers use boards on Pinterest for showcasing their products. However, the characteristics of popular fashion boards are not well-known. These characteristics can be used for predicting popularity of a nascent board. Further, newly formed boards can organize their content in a way similar to the popular fashion boards to garner enhanced popularity. What properties on these fashion boards determine their popularity? Can these properties be systematically quantified? In this paper, we show how social, temporal and image signals can together help in characterizing the popular fashion boards. In particular, we study the sharing/borrowing behavior of pins and the image content characteristics of the fashion boards. We analyze the sharing behavior using social and temporal signals, and propose six novel yet simple metrics: originality score, retention coefficients, production coefficients, inter-copying time, duration of sharing and speed coefficients. We further study the image based content properties by extracting fashion, color and gender terms embedded in the pin images. We observe significant differences across the popular (highly followed or highly ranked by the experts) and the unpopular (less followed) boards. We then use these characteristic features to early predict the popularity of a board and achieve a high correlation of 0.874 with low RMSE value. Our key observation is that likes and repin retention coefficients are the most discriminatory factors of a board's popularity apart from the usage of various color, gender and fashion terms.