CFRS: A Trends-Driven Collaborative Fashion Recommendation System

Abstract

Fashion has a great impact in everyday life and therefore, people pay close attention to the way they dress. Fashion item recommendation is typically a manual, curated process, where experts recommend items and trends to large populations. However, there is increasing use of automated, personalized recommendation systems, which have valuable applications in e-commerce websites. In this paper, we propose a collaborative fashion recommendation system, called CFRS. Apart from classic features, we propose a new metric, called trend score. Trend score shows how trendy a product is and is calculated taking into account the ratings provided by CFRS users (fashion experts and registered users). In particular, users rate (like/ dislike scale) current trends about colors, prints and materials. Finally, trend score is used a) for sorting products of each category from trendiest options to classic ones and b) to recommend trendy products from different clothing categories.