Akademik Veri Yönetim Sistemi
Yüksek A. G. (Yürütücü), Yalçın E., Delibaş E., Şeker A.
Yükseköğretim Kurumları Destekli Proje, 2019 - 2020
With the rapid improvement of Internet
technologies, recommender systems have been widely used. Recommender systems
are specialized in suggesting appropriate items to users with respect to their
personal characteristics and past preferences without requiring any effort of
users. Products/Services usually recommended to individuals, but there are
situations in which a group of individuals collectively participates in a
single activity such as visit a famous restaurant for lunch with their
colleagues, watch a funny TV program with their family, or go to a movie with
their friends. To support the recommendation process in such social activities,
group recommendation has been recently employed effectively. The main goal of a
group recommendation involves providing appropriate information for all members
in a group by analyzing the characterizes and the propensity of the group.
Nevertheless, there are some challenges in the group recommendation. One of the
main challenges in the group recommendation is that how to consider a group as
a whole. Most extant studies use aggregation strategies to determine group
preferences/predictions. An aggregation strategy is an approach that aggregates
individual preferences of group members to recommend items to a group. Although
several aggregation strategies have been developed so far, there is a need for
developing novel aggregation strategies in terms of enhancing group
recommendation’s quality. Identification of group members is another challenge
in a group recommendation system, because groups are usually unknown. In other
words, grouping similar individuals to the same group so that each group
receives the most suitable recommendation is an important task in the group
recommendation. The main topic for this research project is to research how to
create frameworks which will overcome such challenges in the group
recommendation. In order to aggregate predictions/preferences
accurately, we will first try to develop novel aggregation strategies. In doing
so, we will address multi-criteria rating based systems in addition to the
traditional single-criterion based rating systems. After that, we plan to build
an approach to determine the most suitable groups by analyzing the
characteristics of the group members.