Kompe, RalfEmele, MartinGoronzy, SilkeMencl, RobertTorge, SunnaPeissner, MatthiasRöse, Kerstin2017-11-182017-11-182003https://dl.gi.de/handle/20.500.12116/6040In people’s home the amount of on-line available digital content such as video, music, or pictures is rapidly increasing. First home server products are available. Efficient selection of content will be a problem for users. We developed a flexible multimodal dialog system to overcome this. A keycomponent is the user model, which learns very fast from user behaviour. All information like key-words are determined automatically and scored according to their relevance. No manual setting is required.User modelrecommendationmulti-modal dialogText