Over- and underestimation in different product domains

Nava Tintarev, Judith Masthoff

Research output: Chapter in Book/Report/Conference proceedingPublished conference contribution


This paper investigates the effects of over and underestimation on the perceived Effectiveness (helpfulness) of recommender systems. We consider four different product along two dimensions, degree of objectivity and investment. Overestimation was considered more severely than underestimation with regard to perceived Effectiveness. Overestimation was also considered more severely in high investment domains compared to low investment domains. In addition, we surveyed the effect of different gaps between initial (initial impression) and final ratings (true estimate). We found that for gaps which remained in the negative half of the scale were considered less Effective than gaps which crossed over from good to bad (or from
bad to good), and gaps which remained in the positive half of the scale.
Original languageEnglish
Title of host publicationWorkshop on Recommender Systems
Subtitle of host publicationECAI 2008, 18th European Conference on Artificial Intelligence, Proceedings
EditorsMalik Ghallab, Constantine D. Spyropoulos, Nikos Fakotakis, Nikos Avouris
Place of PublicationGreece
PublisherIOS Press
Number of pages6
ISBN (Print)1586038915 , 978-1586038915
Publication statusPublished - 31 Jul 2008
Event18th European Conference on Artificial Intelligence (ECAI 2008) - Patras, Greece
Duration: 21 Jul 200825 Jul 2008


Conference18th European Conference on Artificial Intelligence (ECAI 2008)


  • Recommender systems
  • explanations
  • user studies


Dive into the research topics of 'Over- and underestimation in different product domains'. Together they form a unique fingerprint.

Cite this