An enhanced aggregation method considering deviations for a group recommendation
- Authors
- Seo, Young-Duk; Kim, Young-Gab; Lee, Euijong; Seol, Kwang-Soo; Baik, Doo-Kwon
- Issue Date
- 1-3월-2018
- Publisher
- PERGAMON-ELSEVIER SCIENCE LTD
- Keywords
- Group recommendation; Aggregation method; Upward leveling; Deviation; Average; Approval voting
- Citation
- EXPERT SYSTEMS WITH APPLICATIONS, v.93, pp.299 - 312
- Indexed
- SCIE
SCOPUS
- Journal Title
- EXPERT SYSTEMS WITH APPLICATIONS
- Volume
- 93
- Start Page
- 299
- End Page
- 312
- URI
- https://scholar.korea.ac.kr/handle/2021.sw.korea/76787
- DOI
- 10.1016/j.eswa.2017.10.027
- ISSN
- 0957-4174
- Abstract
- The goal of a group recommendation involves providing appropriate information for all members in a group. Most extant studies use aggregation methods to determine group preferences. An aggregation method is an approach that aggregates individual preferences of group members to recommend items to a group. Previous studies on aggregation methods only consider high averages, counts, and rankings to provide recommendations. However, the most important component of a group recommendation involves ensuring that majority of the members in a group are satisfied with the recommended results. Therefore, it is necessary to consider the deviation as an important element in aggregation methods. The present study involves proposing an upward leveling (UL) aggregation method that considers deviations for group recommendations. The UL recommends items with low deviations and high averages in conjunction with frequency of positive rating counts for group members. Furthermore, the effectiveness of the UL is validated to perform a comparative evaluation with existing aggregation methods by using the normalized discounted cumulative gain (NDCG) and diversity. The results indicate that the UL outperforms all the baselines and that the deviation plays an important role in the aggregation method. (C) 2017 Elsevier Ltd. All rights reserved.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - College of Informatics > Department of Computer Science and Engineering > 1. Journal Articles
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.