Making sense of ratings

Abstract

This paper proposes a common ontology for ratings, i.e. for quantitative user feedback data. Such a framework allows for semantic interoperability of data that adheres to it, which in turn enables the re-use, by making it independent from the original system.In contrast to prior attempts to establish an unambiguous vocabulary, this approach introduces two components that are in our view necessary to formally understand what a user's rating actually means. The first is the aspect or facet, i.e. the viewing angle that was chosen to look at the rated thing. The second is the meta-model of scales following the scales of measurement that are widely used in descriptive statistics. So in plain words, we allow to formally specify how many out of how many score points something gets and with regards to what.We follow the open world assumption of the Web Ontology Language (OWL) and design a vocabulary that is not specific to any domain. In turn, we rely on the premise that all domain specific concepts are available as semantic web resources with appropriate URIs.