Can models of author intention support quality assessment of content?

Academics seek to find, understand and critically review the work of other researchers through published scientific articles. In recent years, the volume of available information has significantly increased, partly due to technological advancements and partly due to pressures on academics to 'publish or perish'. This amount of papers presents a challenge not only for the peer-review process but also for readers, particularly inexperienced readers, to find publications of high quality. Whilst one might rely on citation or journal rankings to help guide this decision, this approach may not be completely reliable due to biased peer-review processes and the fact that the citation count of an article does not per se indicate its quality. Here, we analyse how expected author intentions in a Related Work section can be used to indicate its quality. We show that author intentions can predict the quality with reasonable accuracy and propose that similar approaches could be used in other sections to provide an overall picture of quality. This approach could be useful in supporting peer-review processes and for a reader in prioritising articles to read.