Enhancing Quality Assessment Using Perspective Detection

As humans, we assess Web documents on two dimensions. On the one hand, we evaluate their quality by judging how precise, accurate or neutral the information contained in documents is, or how reliable their sources are based on their reputation. On the other hand, we consider which perspectives are represented in documents. Do the authors present information from their own or someone else’s perspective? How (un)certain are sources about the truth of statements? In turn, perspectivization of information may affect our quality assessments.

In this project, two students will collaborate to improve and enrich the output of an existing tool (Quality Assessment Service) which, at the moment, assesses the quality of Web documents at a very coarse level: documents are rated on a 1-5 likert scale learnt from a set of assessments collected from experts.