A recommender system supporting users of an integrated web-based interface

Abstract

Recommender systems have gained considerable popularity for their helpful suggestions to end-users. However, limited research has been conducted exploring their potential of supporting users as they navigate through integrated interfaces. This research, therefore, focuses on implementing a recommender system as support tool for users dealing with infrequently used complex interfaces. The study was carried out at CashDesk, a company specialising in cash register software (POS) for food delivery and take-away restaurants. The targeted user group included restaurant owners, and the menu editor was selected as a representative complex infrequently used interface. Issues with the menu editor were identified through a customer support log analysis and interviews with CashDesk support employees. Subsequently, three potential recommender system interfaces were prototyped and demonstrated to interviewed customers. Using their feedback, one of the three interfaces has been optimised and evaluated through screenshots and use-cases in a questionnaire. Results indicated a positive reception of the recommender system by respondents, including improved perceived user experience and usability of the menu editor. Users also expected faster menu creation and increased inspiration as additional benefits. While the recommender system is likely to have a positive impact on the menu editor, there is no significant anticipation of a reduction in the need for external support in navigating the interface, as assessed by the participants. However, the recruited subjects primarily identified themselves as highly experienced users. Further research is recommended, involving evaluating an actual recommender system including user interactions in a broader application and user context.