A smart business system for jewellery retailing

Abstract

The Hong Kong jewellery industry is facing challenges in adopting e-commerce, especially for product visualisation, personalised marketing, and building customer faith. Digital technologies, including mixed reality, face recognition, and recommender system, can help to deal with such challenges by improving product visibility, jewellery recommendation, and customer experience. However, the development and implementation of these digital technologies require considerable financial resources and professional expertise which are only affordable by large jewellery enterprises. Small and medium jewellery companies suffer from digital gaps due to their lack of resources and technological support.

To meet these challenges, this thesis proposes a smart business system for the jewellery industry by taking a medium sized jewellery retail company, Company X, as a primary target. The system tackles the identified challenges in Company X by enhancing inventory management, product visualisation, customer information utilisation, promotion quality, communication effectiveness, and data management.

The research work mainly focuses on the development of the modules for mixed reality for virtual jewellery fitting (MRVJF) and personalised marketing (PM), in consideration of time and budget limitations, availability of functionalities, and potential benefits. The MRVJF incorporates mixed reality, face recognition, and hand recognition technologies to improve jewellery visualisation and user-friendliness with interactive and markerless virtual fitting on smartphones. The PM provides customised product promotion by utilising a recommender system to identify customers’ preferences based on their browsing records and purchase histories.

The MRVJF overlays virtual jewellery products onto customers to visualise the fitting effect before manufacture. A positional fine-tuning algorithm is developed to enhance the positional accuracy of virtual products. Jewellery photo shooting requirements are proposed to improve jewellery realisation and reduce the system development cost.

A hybrid jewellery recommender system is developed in the PM to generate jewellery recommendations by integrating content-based filtering and knowledge-based recommender system. The criteria for recommending matching jewellery are identified. Professional jewellery recommendations are provided as reference to salespersons and customers.

The significance of this thesis is to increase the competitiveness of the Hong Kong jewellery retailers in the global commodity chain through technologies. With support of the positional fine-tuning algorithm, the MRVJF integrates and promotes the business of Company X with e-commerce. The jewellery visualisation improvement reduces tailor-made jewellery defective rate, transforms customer shopping habits, and broadens the applications of virtual jewellery fitting from entertainment to facilitate business operations. The PM digitises the expertise of jewellery experts to provide professional jewellery matching recommendations and customised promotion to customers anytime, anywhere.