Situation assessment for non-intrusive recommendation

Abstract

With the rapid growth of mobile applications, the user is increasingly confronted with a lot of information and tend to reject notifications sent by applications installed within his/her mobile device. This rejection affects the performance of many systems, especially proactive recommender systems. Therefore, it is no longer enough for a recommender system to determine what to recommend according to users' needs, but it also has to deal with the risk of disturbing the user during the recommendation process. We believe that the several embedded applications within the user's device along with other parameters could help understand and assess the user's interruptibility in some situations. In this paper, we address intrusiveness within a proactive recommendation approach that makes use of the user's context and the applications embedded within the user's mobile device in order to assess the intrusiveness level of a given situation before recommending.

A probabilistic model for intrusive recommendation assessment

“…The results that we obtained have shown that the proposed approach for intrusiveness assessment performs better when µ equals 1, which is normal since we do not address a large volume of data like it is typically addressed within Information Retrieval tasks. As illustrated in Figure 1, the proposed approach using the situation's features that we considered scores 88.5% for the MAP compared to other features' combinations, to two baseline approaches and to a Case Based Reasoning approach that we proposed in a previous work [2]. Baseline A which sends recommendations without considering the user's interruptibility scores 64% for the MAP.…”

“…Baseline B which consists in not sending a recommendation when an application is ON, scores 50.81%. The CBR approach proposed in [2] consists in using the user's analogous past situations, that are most similar to the actual situation, to gure out if we could interrupt the user's current activity and send a recommendation. The CBR approach scored a MAP of 87% which is slightly less accurate than the proposed probabilistic approach.…”