TY - JOUR
T1 - Social choice-based explanations
T2 - An approach to enhancing fairness and consensus aspects
AU - Tran, Thi Ngoc Trang
AU - Atas, Muesluem
AU - Le, Man Viet
AU - Samer, Ralph
AU - Stettinger, Martin
PY - 2020/3/28
Y1 - 2020/3/28
N2 - Explanations are integrated into recommender systems to give users an insight into the recommendation generation process. Compared to single-user recommender systems, explanations in group recommender systems have further goals. Examples thereof are fairness, which helps to take into account as much as possible group members’ preferences and consensus, which persuades group members to agree on a decision. In this paper, we proposed different types of explanations and found the most effective ones in terms of increasing the fairness perception, consensus perception and satisfaction of group members with regard to group recommendations. We conducted a user study to evaluate the proposed explanations. The results show that explanations which consider the preferences of all or the majority of group members achieve the best results in terms of the mentioned dimensions. Besides, we discovered positive correlations among these aspects. In the context of repeated decisions, group members’ satisfaction from previous decisions are helpful to improve the fairness perception of users concerning group recommendations and speed up the group decision-making process. Furthermore, we found out that gender diversity does influence the perception of users regarding the mentioned dimensions of the explanations. Although the proposed explanations were analyzed in group decision scenarios for non-configurable (no-attribute) items, there exist potential possibilities to apply them to explanations for configurable items.
AB - Explanations are integrated into recommender systems to give users an insight into the recommendation generation process. Compared to single-user recommender systems, explanations in group recommender systems have further goals. Examples thereof are fairness, which helps to take into account as much as possible group members’ preferences and consensus, which persuades group members to agree on a decision. In this paper, we proposed different types of explanations and found the most effective ones in terms of increasing the fairness perception, consensus perception and satisfaction of group members with regard to group recommendations. We conducted a user study to evaluate the proposed explanations. The results show that explanations which consider the preferences of all or the majority of group members achieve the best results in terms of the mentioned dimensions. Besides, we discovered positive correlations among these aspects. In the context of repeated decisions, group members’ satisfaction from previous decisions are helpful to improve the fairness perception of users concerning group recommendations and speed up the group decision-making process. Furthermore, we found out that gender diversity does influence the perception of users regarding the mentioned dimensions of the explanations. Although the proposed explanations were analyzed in group decision scenarios for non-configurable (no-attribute) items, there exist potential possibilities to apply them to explanations for configurable items.
KW - Group composition
KW - Group decision making
KW - Group recommender systems
KW - Preference aggregation strategies
KW - Social aspects
KW - Social choice theory
KW - Social choicebased explanations
UR - http://www.scopus.com/inward/record.url?scp=85083587302&partnerID=8YFLogxK
U2 - 10.3897/jucs.2020.021
DO - 10.3897/jucs.2020.021
M3 - Article
AN - SCOPUS:85083587302
SN - 0948-695X
VL - 26
SP - 402
EP - 431
JO - Journal of Universal Computer Science
JF - Journal of Universal Computer Science
IS - 3
ER -