You are here

Item-Based Explanations for User-Based Recommendations

Authors: 

Marius Kaminskas, Fred Durao, Derek Bridge

Publication Type: 
Refereed Original Article
Abstract: 
Explanations can increase user satisfaction with recommender systems. While it is relatively easy to explain the recommendations of a content-based or an item-based collaborative recommender system, user-based collaborative recommendations are harder to explain. In this work, we adopt an approach from the literature that generates explanation rules for user-based collaborative-filtering recommendations. These rules are item-based: for example, “If you liked Toy Story then you might also like Finding Nemo”. We modify the approach by proposing two new, alternative measures of explanation rule quality. We evaluate the two new measures in a user study and show that users prefer explanation rules whose antecedents are both accurate and unique with respect to the recommended item
Digital Object Identifer (DOI): 
-
Publication Status: 
Published
Publication Date: 
23/03/2017
Journal: 
Procs. of eKNOW 2017, The Ninth International Conference on Information, Process, and Knowledge Management
Pages: 
65-70
Research Group: 
Institution: 
UCC
Open access repository: 
Yes