Interactive construction of personalized fashion capsule wardrobes with alternative item recommendations

In the fashion field, a Capsule Wardrobe, often referred to as CW, is a minimal and essential collection of fashion items from which numerous attractive and diverse coordinations can be created. In this paper, to assist fashion beginners, we propose a framework for semi-automatically constructing personalized CWs. More specifically, by utilizing direct interactions with users and two types of system recommendations on fashion items, the framework interactively constructs high-quality CWs that reflect individual preferences. The results of evaluation experiments conducted on real world dataset indicate that the proposed framework can incorporate user preferences appropriately without decreasing the quality of CWs.