Search costs vs. User satisfaction on mobile

Abstract

Information seeking is an interactive process where users submit search queries, read snippets or click on documents until their information need is satisfied. User cost-benefit models have recently gained popularity to study search behaviour. These models assume that a user gains information at expense of some cost. Primary assumption is that an adept user would maximize gain while minimizing search costs. However, existing work only provides an estimate of user cost or benefit per action, it does not explore how these costs are correlated with user satisfaction. Moreover, parameters of these models are determined by desktop based observational studies. Whether these parameters vary with device is unknown. In this paper we address both problems by studying how these models correlate with user satisfaction and determine parameters on data collected via mobile based search study. Our experiments indicate that several parameters indeed differ in mobile setting and that existing cost functions, when applied to mobile search, do not highly correlate with user satisfaction.