Classical vs. crowdsourcing surveys for eliciting geographic relevance criteria

Abstract

Geographic relevance aims to assess the relevance of physical entities (e.g., shops and museums) in geographic space for a mobile user in a given context, thereby shifting the focus from the digital world (the realm of classical information retrieval) to the physical world. We study the elicitation of geographic relevance criteria by means of both a classical survey and an Amazon Mechanical Turk (a crowdsourcing platform) survey. This allows us to obtain three results: first, we gather a set of criteria and their relative importance; second, we gain a first insight on the differences between geographic relevance and classical relevance as commonly understoon in the IR field; and third we draw some considerations on the agreement, on the importance of specific criteria, among the participants to the classical and the crowdsourcing surveys.