Action Suggestion Using Situation Rules

Nowadays we can see a new era of mobile computing springing up. Mobile devices more often than not provide incomparably more relevant information and context about their users than was ever available on desktops or within classic web browsing. With this a new branch of research for autonomous software is forming. The aim is to recognize usage situations to let an application decide on performing a specific action autonomously. In this paper we describe our novel method for learning users´ situation preferences to suggest the right moments for performing specific actions independently. Such action can be for example autonomous news push in a news application, but it can be just as easily applied to songs or microblog recommenders. User preferences are described with situations the users encounter throughout the time and rules that are based on either implicit or explicit feedback from the users. The focus of this paper is on the action suggestion method for the right moment for performing an action alongside which we also introduce a few usage scenarios, discuss on characteristics and limits of our method and present experiments that evaluate on our method´s performance in various scenarios.