Influencing Factors for User Context in Proactive Mobile Recommenders

Abstract

Proactive recommender systems break the standard request-response pattern of traditional recommenders by pushing item suggestions to the user when the situation seems appropriate. To support proactive recommendations in a mobile scenario, we have developed a two-phase proactivity model based on the current context of the user. In this paper, we explain our approach to model context by identifying different components: user and device status, and user activity. We have conducted an online survey among over 100 users to investigate how different context attributes influence the decision when to generate proactive recommendations. Thus, we were able to acquire appropriateness factors and weights for the context features in our proactivity model.