The pareto frontier of utility models as a framework for evaluating push notification systems

Gaurav Baruah, Jimmy Lin

Abstract: We propose a utility-based framework for the evaluation of push notification systems that monitor document streams for users' topics of interest. Our starting point is that users derive either positive utility (i.e., "gain") or negative utility (i.e., "pain") from consuming system updates. By separately keeping track of these quantities, we can measure system effectiveness in a gain vs. pain tradeoff space. The Pareto Frontier of evaluated systems represents the state of the art: for each system on the frontier, no other system can offer more gain without more pain. Our framework has several advantages: it unifies three previous TREC evaluations, subsumes existing metrics, and provides more insightful analyses. Furthermore, our approach can easily accommodate more refined user models and is extensible to different information-seeking modalities.