Development of personalized fashion recommendations 

Competence CenterInformation Retrieval and Machine Learning
ContactProf. Dr.-Ing. Sahin AlbayrakTill Plumbaum
Partner: Outfittery GmbH

DressRec is a project in cooperation with the Outfittery GmbH for the development of an expert system for the combination of complete outfits based on fitting and personal preferences. DAI-Labor focuses on the development of automated methods for personalized fashion recommendations. 

Outfittery relies on the expertise of fashion stylists to provide personalized compilations of well-matched outfits for their customers. This task requires experience, intuition and knowledge of customers’ preferences. The fashion experts consider a variety of aspects, such as color and material compatibility, fits and current trends to collate clothes that best match the company’s individual customers’ preferences. In this web-based business model, fashion advice is provided with the help of an electronic questionnaire and a personal phone call. However, following this method, understanding the customers’ preferences is rather limited. Fashion specialists create individual outfits based on this limited knowledge, which are then sent to customers. Customers check the outfits and decide what to keep and what to send back. The economic aim of this project is to reduce the return rate, hence minimizing shipping and effort costs. 

DAI-Labor will primarily provide two scientific contributions:

(1) Innovative algorithms to narrow down the amount of matching clothes for a customer based on the user profile and expert knowledge.

(2) Algorithms for continuous learning (parameter configurations) are researched, which adapt recommendation strategies with respect to a target variable, e.g. reducing the return rate. 

The DressRec project is a cooperation between DAI-Labor and Outfittery GmbH and is funded by the Federal Ministry of Economics and Technology (BMWi).