=Paper=
{{Paper
|id=Vol-1618/Poster2
|storemode=property
|title=Supporting Group Decision Making with Recommendations and Explanations
|pdfUrl=https://ceur-ws.org/Vol-1618/Poster2.pdf
|volume=Vol-1618
|authors=Thuy Ngoc Nguyen,Francesco Ricci
|dblpUrl=https://dblp.org/rec/conf/um/NguyenR16
}}
==Supporting Group Decision Making with Recommendations and Explanations==
https://ceur-ws.org/Vol-1618/Poster2.pdf
Supporting Group Decision Making with
Recommendations and Explanations
Thuy Ngoc Nguyen Francesco Ricci
Free University of Bozen-Bolzano, Free University of Bozen-Bolzano,
Piazza Domenicani 3, Piazza Domenicani 3,
Bolzano, Italy Bolzano, Italy
ngoc.nguyen@unibz.it fricci@unibz.it
ABSTRACT mobile system that supports the process of making decision
In this poster we illustrate the interaction design of a mo- in groups. More concretely, the system aims at supporting
bile system that facilitates group decision making by allow- tasks that group members are likely to undertake during
ing group members to be engaged in a discussion which is the decision process such as asking for information, making
actively supported by recommendation functions and expla- comparisons, or seeking a rationale for options.
nations. The interactions between the users and the system
are monitored in order to proactively offer appropriate di- 2. INTERACTION WITH THE SYSTEM
rections and suggestions. Unlike many state of the art group
The motivation underlying our interaction design is de-
recommenders, which solely mediates users’ preferences and
rived from studies on the functional theory of group de-
suggest items that are likely to be acceptable by all the group
cision making which suggest that it is structured in four
members, our system acts as a facilitator that guides and
stages: Orientation - Discussion - Decision - Implementa-
helps the group members in coming up with an agreement
tion (ODDI model) [3]. Furthermore, following the indica-
and a final decision.
tion that decision makers often seek and construct reasons
in order to resolve the conflict and justify their choice when
Keywords they are faced with the need to choose [7], our system aims
Group Recommender Systems; Group Decision Support. at supporting the decision process by providing explanations
for all the generated recommendations and suggestions.
In the following paragraphs, we describe a typical inter-
1. INTRODUCTION action with our system called STSGroup (South Tyrol Sug-
Recommending items to a group has been usually seen as gests for Group), and we show some of its primary functions.
a complicated function due to the fact that conflicting pref- STSGroup is an Android-based mobile application that ex-
erences between group members can easily arise. Moreover, tends to groups STS [1], a context-aware places of interest
group members often change their mind in an unpredictable (POIs) recommender originally devoted to individuals. Let
way while interacting with each others and the system [5]. us assume a tourist or a citizen is looking for a POI (in
The research in group recommender systems (GRSs) has South Tyrol, Italy) for her group to visit together. With
already seen some contributions in which the role of the the support of the system, the user is able to:
interaction between users and system has been recognized Make companions: this function allows the user to spec-
as important for the group members to reach a consensus. ify her companions through appropriate system screens in-
For instance, the critiquing technique clearly exemplifies cluding: searching companions by user name, sending con-
this direction, and it is often implemented in naturalistic nection requests and tagging companions. Once a group of
negotiations. Specifically, in Collaborative Advisory Travel people that are connected by the “companion” relation wants
System critiquing is used for allowing each group member to visit a POI, the discussion is ready to start. Note that
to send a “critique” to the other members, thereby sharing users can always access functions that are already available
thoughts about a specific option [6], and Where2eat intro- in STS; for instance they can specify context variables such
duced interactive multi-party critiquing which is an exten- as their mood, or browse their (individually) personalized
sion of the critiquing concept to a computer-mediated con- recommendations.
versation between individuals [4]. Recently, a group deci- Start a discussion: one user (in a group) can autonomously
sion support environment Choicla has been developed that search and propose items that are thought to be suitable for
allows the flexible definition of decision functionality in a her group of companions. A discussion session is started
domain-independent setting [8], [9]. as soon as a first item proposal is sent to the other group
However, in the context of group recommendation still members. The other group members can then browse this
not enough attention has been devoted to understand how proposal and add some others on their own.
the process of making choices in groups can be supported Evaluate proposals: all proposed items are moved into
[2]. In fact, social scientists studying group dynamics have the group discussion space displayed in a news feed, where
stressed the importance of various aspects and steps of the group members can react to them by rating them as: likes
full decision process adopted by a group in determining the (thumb up), dislikes (thumb down), or favorites (heart icon).
quality of the output decision [3]. Motivated by these find- User can also tag proposals with comments and emoticons
ings, we therefore here introduce the interaction design of a (see Figure 1a). A summary comparison panel aggregating
Figure 1: Screenshots of STSGroup, from left to right: (a) Group discussion, (b) New item recommendations
for a group, (c) Hint suggestion, and (d) Final choice suggestion.
the members’ likes, dislikes and favorites is always shown on the whole decision process and help group members under-
the top of the feed in order to keep users aware of the other stand each others. The research is still in progress. We are
members’ preferences. The panel is updated automatically currently implementing the recommendation algorithms and
when there is any change in the preferences expressed by we will conduct a live user study to evaluate the effectiveness
any group member. of our design.
Ask/get recommendations for new items: during the
discussion, in case a user would like to see some more pro- 4. REFERENCES
posals, in addition to those already made, she can ask for [1] M. Braunhofer, M. Elahi, and F. Ricci. Usability
new item recommendations (see Figure 1b). The system assessment of a context-aware and personality-based
can also proactively propose new items when it detects that mobile recommender system. E-commerce and web
this could be valuable: for instance when users change often technologies, pages 77–88, 2014.
preferences for items, showing that they are unsure about [2] L. Chen, M. de Gemmis, A. Felfernig, P. Lops, F. Ricci,
the current proposals (see Figure 1c). Recommendations and G. Semeraro. Human decision making and
are augmented with explanations that are computed on the recommender systems. ACM Transactions on
base of the group members’ actions and contexts and a ra- Interactive Intelligent Systems (TiiS), 3(3):17, 2013.
tionale for the system recommendations is given. Recom- [3] D. Forsyth. Group Dynamics. Wadsworth Cengage
mendations take into account the discussion and the role of Learning, 6th edition, 2014.
users. For example, the more items a user rates, the higher
[4] F. Guzzi, F. Ricci, and R. Burke. Interactive
the importance she will have in the preference aggregation
multi-party critiquing for group recommendation. In
step of the recommendation computation. We also assign a
Proceedings of the 5th ACM Conference on RecSys’11,
higher importance to users who are in somewhat vulnerable
pages 265–268, IL, USA, 2011.
contexts such as bad mood, depression, or tiredness. This
means that items similar to what they have proposed are [5] J. Masthoff. Group recommender systems: aggregation,
more likely to appear in the recommendation list. satisfaction and group attributes. In F. Ricci,
Hints: hints are supplementary information about items, L. Rokach, and B. Shapira, editors, Recommender
which are added automatically by the system to the flow of Systems Handbook, pages 743–776. Springer, 2015.
the comments, or suggestions for better using some of the [6] K. McCarthy, L. McGinty, B. Smyth, and M. Salamo.
system functions. The needs of the many: a case-based group
Ask for a choice: when facing difficulties in arriving to recommender system. Advances in Case-Based
a final decision, the user can refer to the choice suggestion Reasoning, pages 196–210, 2006.
function (see Figure 1d). At this point the system invokes [7] E. Shafir, I. Simonson, and A. Tversky. Reason-based
a preference aggregations strategy, such as Majority Vote, choice. Cognition, 49(1-2):11–36, 1993.
and all the proposed items are ranked with respect to it. [8] M. Stettinger and A. Felfernig. Choicla: Intelligent
Explanations are also constructed based on this strategy. decision support for groups of users in the context of
personnel decisions. In Proceedings of the ACM RecSys’
14 IntRS Workshop, pages 28–32, 2014.
3. CONCLUSIONS [9] M. Stettinger, A. Felfernig, G. Leitner, S. Reiterer, and
In this poster, we have described the interaction design of M. Jeran. Counteracting serial position effects in the
a new mobile recommender system that supports decision choicla group decision support environment. In
making in groups by offering a variety of recommendation Proceedings of the 20th International Conference on
and explanation functions. We have argued that, in order to Intelligent User Interfaces, pages 148–157, GA, USA,
make a better decision in groups, the system should support 2015.