=Paper=
{{Paper
|id=Vol-2537/paper-08
|storemode=property
|title=TindArt, an Application to Understand Cultural Tastes
|pdfUrl=https://ceur-ws.org/Vol-2537/paper-08.pdf
|volume=Vol-2537
|authors=Daniel Zilio
|dblpUrl=https://dblp.org/rec/conf/fdia/Zilio19
}}
==TindArt, an Application to Understand Cultural Tastes==
https://ceur-ws.org/Vol-2537/paper-08.pdf
TindArt, an application to understand cultural
tastes
Daniel Zilio1[0000−0001−7107−8858]
Department of Cultural Heritage – University of Padua
Piazza Capitaniato, 7 – 35139 Padua – Italy
{daniel.zilio.3}@phd.unipd.it
Abstract. In this paper an Android application called TindArt is pre-
sented. It has been developed to investigate a way to profile the user in
cultural contexts, with future applications to Recommender Systems for
museum visits. The purpose of this research also includes the study of
the User Experience with this application to understand how it could be
used in a real museum context. Two pilot studies are also presented.
Keywords: Recommender System · User Profiling · Cultural Heritage
· Museum · User Experience
1 Introduction and Motivation
One of the most important institutions in our society is the museum. Since
its foundation this entity made a fundamental contribution to the preservation,
conservation and communication of Cultural Heritage. Inside museums we can
find artworks, ruins of the human and natural past, and many other examples
of items that are the bearers of human knowledge.
There are many museums in the world, in Italy alone we can enumerate nearly
five thousand1 of them. Despite this, Italy does not have a museum on the top ten
visited in the world2 , although this country is one of the most popular tourism
destinations. There are many reasons for this discrepancy, one of these is that
museums are not very attractive to tourists and citizens, and unfortunately they
are seen as boring places3 . This perception might be caused by the problem of
the transmission of the cultural message that is intrinsic to the objects inside,
especially for art and archaeological museums. In general we find panels with,
Copyright c 2019 for this paper by its authors. Use permitted under Creative Com-
mons License Attribution 4.0 International (CC BY 4.0). FDIA 2019, 17-18 July
2019, Milan, Italy.
1
ISTAT, I musei, le aree archeologiche e i monumenti in Italia. Anno 2015.
2
Global Attractions Attendance Report, Themed Entertainment Association (TEA)
and the Economics practice at AECOM, 2018, https://www.aecom.com/content/
wp-content/uploads/2019/05/Theme-Index-2018-5-1.pdf
3
European Report CULTURAL ACCESS AND PARTICIPATION, 2013: http://
ec.europa.eu/commfrontoffice/publicopinion/archives/ebs/ebs_399_en.pdf
too often, walls of text, as well as museum layouts that are easily understood by
experts but not for the average visitor. To tackle these problems, I would like to
reduce this gap by connecting the visitor’s necessities with the museum entity.
The instrument with witch I choose to do so is the Recommender System.
Recommender Systems (RSs) are powerful tools to understand the tastes
and the profile of a user and, using this information, we can suggest to him the
item from a set of items [1] [9]. Recommender Systems are used in several areas,
in particular in the commercial field, but they can be used in many different
situations. Amazon.com [6], that suggests us what we can buy, is one of the
main examples, Netflix and Facebook are others.
In this paper, I propose a way to use these techniques to create personalised
museum visits, starting with the problem of profiling users from a cultural stand-
point. The general idea is to use RSs to adapt the museum for the user without
losing the content expressed by the museum.
The concept of visit is not limited to visitor’s itinerary inside a museum but it
also includes the interaction with the items on display, the manner of transmit-
ting the cultural message to the visitor, and so on.
For this scope I developed an Android application, called TindArt, to collect in-
formation about users and an another application to test it. I do not want only
to try to understand the single user profile; another objective is to investigate
the possibility of classifying users, which could be used as a starting point for
personalising museum visits. In addition to collecting information about user
preferences, I collect data about the user experience with TindArt.
The paper is organised as follow: Section 2 is a short corpus of examples of ap-
plications of RSs in museums, Section 3 is a description of the application and
its use, and Section 4 contains the conclusions and proposed avenues of future
work.
2 Related Works
The use of personalised, mostly mobile, museum guides to improve visitor expe-
rience [8] and, consequently attract new visitors [11], are used in an increasing
number of museums around the world. There are different approaches to captur-
ing visitor’s preferences and tastes: asking them directly [3], using them indirectly
during or after a visit to an exhibition [2] [5]. In general the applications of Rec-
ommender Systems to museum contexts are important case of studies [12] [10] [4]
and also their impact on the museum entity is of interest [7].
3 An application for all tastes
The motivations about the realisation of the mobile application TindArt are
twofold: the first one is to create an instrument that can be used from profile
users from a cultural point of view, and the second one is to study the interaction
and the experience of users with this type of instrument, with the intent to use
it in a museum pre-visit 4 .
Although these two aims appear different, it is important for my research to
investigate not only if this method will be a good instrument to obtain the
user’s cultural profile, and generate relevant suggestions, but also to investigate
if this instrument will be really applicable to a real situation as well. This section
is therefore divided into three parts: in Section 3.1 I describe how the app works
and what kind of data it gathers, in Section 3.2 I presents two pilot studies, and
in Section 3.3 I list the specific information collected to analyse user behaviour.
3.1 How TindArt works
TindArt is developed for the Android system and it can be downloaded, at the
date of this publication, in beta version from Google Play5 . The application
mimics the famous app Tinder6 by using a swipe 7 gesture to express preferences
regarding not a person, but in this case, an artwork. The goal is to give the user
an instrument that is easy to use and fairly well-known.
After downloading it, the user creates an account using an email8 , and so a
session is created. In the main screen (see Figure 1.b) we find four implicit
buttons and one not implicit button:
– Logout: used to disconnect from actual session;
– Guida: a small tutorial of the application;
– Progetto: shows information about the project;
– Inizia: starts the artworks evaluation;
– PERSONAL INFORMATION (not implicit): the logo in the centre of the
upper task bar, only used for the pilot study described in 3.2
In Figure 1(a) we can see the main screen of the application. The application
randomly shows an artwork from a set of artworks previously selected, taken
from The Met’s Collections site9 and from The Met’s Heilbrunn Timeline of Art
History10 (more details about this choice are in sections 3.2), and user can only
rate it positively or negatively. There are two ways to generate the preference:
– using the green button for Like or the red button for Nope;
– using the swipe gesture, left to right for Like, right to left for Nope.
There is also a button in the upper right corner named Chat, but it is reserved
for future experiments.
4
Pre-visit is the time between the choice to go in museum and the visit.
5
https://play.google.com/store/apps/details?id=tindar_evo.meeple.tindart
6
https://play.google.com/store/apps/details?id=com.tinder
7
It represents a linear motion of fingers to a screen in order to move onto the next
page, choose something, etc.
8
The user account management is relegated to Google Firebase, https://firebase.
google.com/
9
https://www.metmuseum.org/art/collection/
10
https://www.metmuseum.org/toah/
(a) Home (b) Select
Fig. 1: Home and Main screen of TindArt
A rating session can be stopped at anytime by the user and it is possible to
restart the session a second time. An artwork can be rated more than one time
but the total number of artworks are not known by user and when he completes
all available choices a farewell screen is shown. At the time of this publications
the total number of artworks loaded is 352.
3.2 TindArt for user profiling
As I wrote supra one of the main objectives of TindArt is to collect information
about user’s cultural preferences and, starting from this data, developing a Rec-
ommender System for museum visits.
At the end of the two pilot studies described below I will start considering two
main routes to follow: the first one is create a typical Neighbourhood-based
Collaborative Filtering RS, in which users are compared with a similarity func-
tion. In the second one I would like to use the classification methods to group
users by their cultural tastes and from that, understand if it is possible to make
suggestions not for a single user but for a class of users.
Pilot study without application The first pilot study was conducted before
publishing TindArt. 61 undergraduate students in the course in History and
Conservation of the Artistic and Musical Heritage each chose five artworks from
the above-mentioned website, The Met’s Heilbrunn Timeline of Art History. The
choice of this archive is due to the fact that it is well-realised, and moreover,
its metadata is uniform facilitating database entry. In my research group we are
going to first create a user classification based on the artworks selected by the
students. The idea is to group several art movements into smaller subsets to
understand if students selected artwork from same art movements, and starting
from there, to classify the users. The analyses are still ongoing and the results
will be presented in future publications.
Pilot study The second pilot study is about the collection of user preferences
using TindArt. A group of undergraduate students in Design and Management of
Cultural Tourism 11 are using the application and storing all information gathered
from the app in a database. The idea is to collect user preferences and use these to
make an RS for the visit to a villa-museum situated in Abano Terme (Padua).
Another application, called Ospiti in Villa Bassi 12 , designed to explore this
location was developed by me and my research group and will be upgraded for
this scope.
On the other hand, TindArt is available to a large set of users13 and all
information that will be gathered from it will be used for my research.
3.3 User Experience of TindArt
Another important aspect of this research is to understand if an application like
TindArt could be used in a real museum context and, at the same time, could
investigate the behaviour of the user when he using this tool. To determine
whether the app can be use in real context a set of variables is stored in a
database during the application use. These variables are reported for every single
rated artwork:
– Swipe: a boolean flag that represents if the user rated using the buttons or
the swipe gesture.
– Date: the information about the date could be used to analyse the number of
votes that a user gives in a specific period of time. It also could be a measure
of how many times a user uses the app. This information is a reliable marker
of user decision making.
– Time of choice: the timing expended by the user to vote.
– Resolution of smartphone: I included this passive information to investigate
if different devices could influence the user’s choice.
The number of artworks for which a user gives his preferences before dismissing
the application is another marker that will be considered.
4 Conclusion and future works
In this paper I described an Android application called TindArt with the intent
to develop a tool that could be a base for the creation of a Recommender Sys-
tem for museum visits. The two pilot studies are ongoing and the information
obtained from these will be an important index for this research. For the user pro-
filing problem the next steps will be the application of classification techniques:
the first candidate is the Bayesian Machine Learning methods. Simultaneously
11
The application is available for everybody in Play Store but I’m promoting it espe-
cially to students.
12
https://play.google.com/store/apps/details?id=com.meeple.villabano
13
The applications is available for 2000 users for the beta version.
another Recommender System-oriented method will be analysed to tackle the
development of a museum visit RS, in particular our attention will be focused on
item-based RS methods. The other important aspect discussed is user behaviours
when he using the application, which could give considerable information about
his profile, as well as the possibility to use an application like TindArt in real
situation. As previously written, a case of study in a villa-museum is scheduled.
References
1. Aggarwal, C.C.: Recommender Systems: The Textbook. Springer Publishing Com-
pany, Incorporated, 1st edn. (2016)
2. Antoniou, A., Katifori, A., Roussou, M., Vayanou, M., Karvounis, M., Kyriakidi,
M., Pujol-Tost, L.: Capturing the visitor profile for a personalized mobile museum
experience: an indirect approach (2016), http://eprints.gla.ac.uk/143234/
3. Aroyo, L., Brussee, R., Rutledge, L., Gorgels, P., Stash, N., Wang, Y.: Person-
alized Museum Experience: The Rijksmuseum Use Case. In: Museums and the
Web 2007. San Francisco, USA (April 11-14, 2007), http://www.archimuse.com/
mw2007/papers/aroyo/aroyo.html
4. Berre, D.L., Marquis, P., Roussel, S.: Planning personalised museum visits. In:
Proceedings of the Twenty-Third International Conference on Automated Planning
and Scheduling, ICAPS 2013, Rome, Italy, June 10-14, 2013 (2013), http://www.
aaai.org/ocs/index.php/ICAPS/ICAPS13/paper/view/6025
5. Keller, I., Viennet, E.: Recommender systems for museums: Evaluation on a real
dataset. In: Fifth International Conference on Advances in Information Mining and
Management (Jul 2015)
6. Linden, G., Smith, B., York, J.: Amazon.com recommendations. item-to-item col-
laborative filtering. IEEE Internet Computing 7(1), 76–80 (2003)
7. Loboda, O., Nyhan, J., Mahony, S., Romano, D.: Towards evaluating the impact
of recommender systems on visitor experience in physical museums (09 2018)
8. Lykourentzou, I., X, C., Naudet, Y., Tobias, E., Antoniou, A., Lepouras, G., Vas-
silakis, C.: Improving museum visitors’ quality of experience through intelligent
recommendations: A visiting style-based approach (07 2013)
9. Ricci, F., Rokach, L., Shapira, B., Kantor, P.B.: Recommender Systems Handbook.
Springer-Verlag New York, Inc., New York, NY, USA, 1st edn. (2010)
10. Rossi, S., Barile, F., Galdi, C., Russo, L.: Recommendation in museums: paths, se-
quences, and group satisfaction maximization. Multimedia Tools Appl. 76(24),
26031–26055 (2017). https://doi.org/10.1007/s11042-017-4869-5, https://doi.
org/10.1007/s11042-017-4869-5
11. Tan, E., Oinonen, K.: Personalising content presentation in museum exhibitions
- a case study. Virtual Systems and MultiMedia, International Conference on 0,
232–238 (09 2009). https://doi.org/10.1109/VSMM.2009.42
12. Tesoriero, R., Gallud, J.A., Lozano, M.D., Penichet, V.M.R.: Enhancing vis-
itors’ experience in art museums using mobile technologies. Information Sys-
tems Frontiers 16(2), 303–327 (2014). https://doi.org/10.1007/s10796-012-9345-1,
https://doi.org/10.1007/s10796-012-9345-1