Home > Research > Publications & Outputs > Diversity as a Response to User Preference Unce...

Links

Text available via DOI:

View graph of relations

Diversity as a Response to User Preference Uncertainty

Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSNChapter

Published
Publication date04/2018
Host publicationStatistical Data Science
EditorsNiall Adams, Edwards Cohen
PublisherWorld Scientific
Pages55-68
Number of pages14
ISBN (Electronic)9781786345417
ISBN (Print)9781786345394
Original languageEnglish

Abstract

This chapter considers the problem of choosing a set of website elements to present to a user. An often desirable property of such a set is that it is diverse, i.e. the elements are not all similar to one another. Often this is presented as being a separate objective from that of choosing elements that match the user in some way and which are therefore more likely to be clicked. We present a range of simple and intuitive models based on uncertainty about user preferences that show how diversity emerges naturally as a result of seeking to maximise the probability that the user will click on an element. As such we give an argument as to why diversity is desirable which avoids the need for it as a separate objective. The exact model used affects the diversity of sets chosen as well as the likelihood that the user will click on an element.