A group recommendation system with a multi-modal user interface

This paper describes a proposed recommendation system and multi-modal interface. In this study, the system aims to provide a solution to need for people with specific skills by recommending dependable people having both needed knowledge and discussion skills. The user has opportunities to quickly reach effective solutions through discussion with people recommended by this system. A procedure using a multi-agent system uses data obtained through a social network represented as architecture of relations. The recommendation system in this study is implemented in this algorithm, supplemented with additional information to describe the comprehensive level of trust between members of the discussion, ratings which are retrieved from those member´s evaluations. For some situations, for example, when the user does not agree with the recommendation and wants to update or alter the recommended data, a function has been implemented with which the user can readily coordinate groups. As a way of providing complex information, the multimodal interface represents them with a haptic metaphor for easy understanding and quick manipulation. The use of haptic force feedback displays in the virtual space makes it possible for the user to understand more easily than methods using only the visual modality.