The influence of personal values on music taste: towards value-based music recommendations

Abstract

The field of recommender systems has a lot to gain from the field of psychology. Indeed, many psychology researchers have investigated relations between models that describe humans and consumption preferences. One example of this is personality, which has been shown to be a valid construct to describe people. As a consequence, personality-based recommenders have already proven to be a lead toward improving recommendations, by adapting them to their users' traits.
Beyond personality, there are more ways to describe a person's identity. One of these ways is to consider personal values: what is important for the users in life at the most abstract level. Being complementary to personality traits, values may give another lead towards better user understanding. In this paper, we investigate this, taking music as a use case. We use a marketing interview technique to elicit 22 users' personal values connected to their musical preferences. We show that personal values indeed play a role in people's music preferences, and are the first to propose a map linking personal values to music preferences. We see this map as a first step in devising a value-based user model for music recommender systems.

Full text

Permission to make digital or hard copies of part or all of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for prot or commercial advantage and that copies bear this notice and the full citation
on the rst page. Copyrights for third-party components of this work must be honored.
For all other uses, contact the owner/author(s).
RecSys ’19, September 16–20, 2019, Copenhagen, Denmark
© 2019 Copyright held by the owner/author(s).
ACM ISBN 978-1-4503-6243-6/19/09.
https://doi.org/10.1145/3298689.3347021
The Influence of Personal Values on Music Taste: Towards
Value-Based Music Recommendations
Sandy Manolios
s.manolios@tudelft.nl
Delft University of Technology
Delft, The Netherlands
Alan Hanjalic
a.hanjalic@tudelft.nl
Delft University of Technology
Delft, The Netherlands
Cynthia C. S. Liem
c.c.s.liem@tudelft.nl
Delft University of Technology
Delft, The Netherlands
ABSTRACT
The eld of recommender systems has a lot to gain from the eld of
psychology. Indeed, many psychology researchers have investigated
relations between models that describe humans and consumption
preferences. One example of this is personality, which has been
shown to be a valid construct to describe people. As a consequence,
personality-based recommenders have already proven to be a lead
toward improving recommendations, by adapting them to their
users’ traits.
Beyond personality, there are more ways to describe a person’s
identity. One of these ways is to consider personal values: what
is important for the users in life at the most abstract level. Being
complementary to personality traits, values may give another lead
towards better user understanding. In this paper, we investigate
this, taking music as a use case. We use a marketing interview
technique to elicit 22 users’ personal values connected to their
musical preferences. We show that personal values indeed play
a role in people’s music preferences, and are the rst to propose
a map linking personal values to music preferences. We see this
map as a rst step in devising a value-based user model for music
recommender systems.
CCS CONCEPTS
•Social and professional topics → User characteristics.
KEYWORDS
Personal values; Musical Taste; Music Recommendations; Psychol-
ogy of Music; User Modelling; Novel Applications; Explainable
Recommender Systems
ACM Reference Format:
Sandy Manolios, Alan Hanjalic, and Cynthia C. S. Liem. 2019. The Inuence
of Personal Values on Music Taste: Towards Value-Based Music Recommen-
dations. In Thirteenth ACM Conference on Recommender Systems (RecSys
’19), September 16–20, 2019, Copenhagen, Denmark. ACM, New York, NY,
USA, 5 pages. https://doi.org/10.1145/3298689.3347021
1 INTRODUCTION
In
right items for the right users. In this, it is useful to exploit knowl
edge of relations between user taste and preferences on the one
recommender systems, it is important for a system to surface the
-
side and various human factors on the other side, which typically
are studied in psychology [
15
]. Indeed, the eld has seen success-
ful incorporation of psychological models into recommender sys-
tems, most notably in personality-based recommender systems [
11
].
However, more factors beyond personality may inuence and drive
user’s preferences.
In this work, we consider another category of factors that have
been validated in psychological literature: the personal values of
a user. Personal values can be dened by what we strive for, and
what is important for us in life, at the most abstract conceptual level.
Where personality can be considered a description of who people
are at the present moment, personal values form a complement to
this, describing who people want to be in the future[25].
As personal values consider an abstract conceptual level, peo-
ple with dierent taste preferences may ultimately share the same
values. Since personal values also are more oriented towards a de-
sired future state, it can be expected that people are intrinsically
motivated towards them. As a consequence, we believe that aware-
ness of personal values may be of interest for recommender system
scenarios, especially in scenarios that strive to broaden a user’s
horizon beyond their regular taste comfort zone. Going beyond this
comfort zone will challenge the user, and may contradict what a
user would intuitively prefer. Yet, if a more challenging recommen-
dation could be related to the user’s personal values, it still may be
explainable and acceptable to the user, even though at rst sight, it
may not be what a user would have expected.
To the best of our knowledge, no concrete value-based recom-
mender systems exist yet. Before realizing any such system, it
should rst be investigated whether the consideration of personal
values in relation to user consumption preference would indeed
make sense. In this paper, we take a rst step towards this, investi-
gating relationships between music preference and personal values.
Employing the laddering technique, which is a well-known and
well-proven interview technique in the eld of marketing, we elicit
personal values guiding music preferences of listeners. From the
qualitative interview data, a hierarchical value map of music prefer-
ence is built, giving useful rst insights into connections to further
investigate in future quantitative recommender-oriented studies.
2 RELATED WORK
2.1 Music taste in psychological literature
In the eld of music psychology, personal factors connected to
musical taste have extensively been studied [
15
]. Age for example
plays a critical role in the development of music taste [
10
,
18
]. As
a consequence, we tend to have preference for the music that was
popular at this phase of our lives [
2
]. Another factor is the inuence
of peers and parents [14], either because of exposure to the music
501

RecSys ’19, September 16–20, 2019, Copenhagen, Denmark Manolios, Hanjalic & Liem, 2019
they prefer, or because users, especially teenagers, tend to use their
musical tastes as “social badges", expressing belonging to their so-
cial group [
15
,
17
]. Education level can also be a predictive factor
of music taste: people with higher education are inclined more
than the others to like “highbrow" music genres, such as classical
music [
19
] and to be “cultural omnivores" (consume a wide range
of diverse cultural items) [12, 34, 35].
2.2 Personality in music preference and
recommender systems
Recent development on personality-based recommender systems
[
1
,
11
,
33
,
36
] show that adding personality information increases
the accuracy of recommendations and the performance of the sys-
tem. Furthermore, personality-based recommendations can be use-
ful in solving the cold-start problem [
15
,
32
,
36
]. In view of this,
many researchers investigated the link between music taste and
personality. [
28
] presents a meta-analysis of 28 articles on this sub-
ject. The main conclusion was that personality is not a reliable
factor by itself to deduct musical preferences in general. However,
some weak, but consistent and signicant correlations have been
shown between certain personality traits and dimensions of music
preference [
3
–
5
,
7
,
8
,
13
,
21
,
28
]. Beside determining what to rec-
ommend, personality can also support the system into determining
how to recommend by adapting its interface to the user [6] or the
degree of diversity of the recommendations [31, 36].
2.3 Personal values
Personal values have been proposed as another psychological model
to describe people. Several models have been proposed and val-
idated; most notably, the models of Rokeach [
26
] and Schwartz
[
29
,
30
]. The Rokeach classication considers 36 personal values,
that fall into 2 categories (Terminal and Instrumental values). The
Schwartz classication considers 10 personal values, that fall into
5 categories: Conservation (caring about one’s safety in every as-
pects of one’s life), Openness to Change (caring about indepen-
dence and discovery), Self-Transcendence (caring for the world),
Self-Enhancement (caring for oneself), and Hedonism.
[
25
] investigated the relationship between personal values and
personality traits. While correlations were found between them,
the conclusion was that the constructs still are “conceptually and
empirically distinct”, thus making personal values an interesting
psychological model to consider besides personality.
2.4 Values in user preference and
recommender systems
The link between personal values and consumption preferences
has been studied in the marketing eld, based on the Means-End
Theory [
9
]. According to this theory, people choose to consume
certain products instead of others, guided by their personal values.
However, to the best of our knowledge, very little to almost no
work has been done to investigate the link between personal values
and recommendations. While “Values and Beliefs" are mentioned
in [
15
], the values mentioned in this work refer to religious and
political orientation, rather than personal values as a psychological
model. While [
22
] considered both personal values and musical
taste, the relation between these was only reported through the
prism of validity of stereotypes, thus putting the main focus on
personal values as perceived by others, rather than personal values
identied by users themselves.
3 STUDY
To investigate which personal values underlie the music taste of
people, and whether some personal values may be more inuential
on music taste than others, we conducted an exploratory qualitative
study. We rst describe the laddering technique, the way in which
this inspired our interviewing methodology, and details of our
interview participants.
3.1 The Laddering Technique
We conduct our qualitative study by employing the laddering tech-
nique [
23
], which is based on the Means-End Theory [
9
]. It is widely
used in the marketing eld [
24
], but has already been applied in
other studies focusing on consumers [
16
,
20
]. The laddering tech-
nique is designed to uncover the personal values that underlie
people’s preferences and consuming behavior. In its rst step, the
technique elicits product attributes from the interviewees: concrete
characteristics of a product, that the interviewee nds important.
Then, for each attribute, the interviewer encourages the respon-
dents to reect critically about the consequences of those attributes,
indicating what makes a specic attribute important for them. For
each of the mentioned consequences, the interviewer will encour-
age the respondent to reect on why this consequence is important,
which may lead to the identication of underlying values that drove
the consumer preference at the most abstract level.
As a consequence, from each interview, multiple ladders will
emerge, that always will start with an attribute, and subsequently
can lead to consequences, followed by values. Based on the fre-
quency of occurrence of connections between elements, the ladders
can then be combined into a Hierarchical Value Map, giving an
aggregated overview of participants’ personal values guiding their
preferences for certain product attributes [23, 24, 27].
3.2 Methodology
Our study was conducted using in-depth face-to-face individual
interviews. The interviews lasted around an hour, but the duration
was often extended to allow the participants to formulate their
answers as elaborately as they wanted. We used the laddering
technique to conduct the interviews, but for the purpose of this
study, the original methodology was slightly adapted to t it to
our research context. Specically, instead of eliciting the product
features that are important for the interviewees, we asked what
kind of music they liked (e.g.,“When it comes to your favourite
music, what are the rst ve things that come to your mind ?"). We
kept the question vague on purpose and emphasized the freedom
in formulating the answer. In doing so, we were also curious about
what type of attributes people thought to be the most appropriate
to describe their music tastes. We choose to investigate people self-
declared preferences instead of their actual consumption because
the former should be more inuenced by people’s personal values
and represent better what is important to them.
502

The Influence of Personal Values on Music Taste RecSys ’19, September 16–20, 2019, Copenhagen, Denmark
Table 1: The dierent categories emerging from our inter-
view studies and their frequencies. The “Total Number of
Occurrences" gives the total number of times this category
appeared in the data; the “Number of Participants mention-
ing this category" only counts one occurrence per intervie-
wee.
Employing insights from three independent coders, we classied
the attributes and consequences mentioned in the ladders into
broader categories suitable for our analysis and ended up with 18
attributes and 18 consequences.
We also identied personal values at the top of some ladders.
Then we used Schwartz’ broad classication [
29
] to which we added
Art and Social Connection as additional values specic to the music
consumption context, for a total of 8 personal values as a basis
to categorize the values found in the interviews. A full overview
of all categories used for analysis, together with their occurrence
frequencies, is given in Table 1.
To build the Hierarchical Value Map, we aggregated all results
into two symmetric matrices, computing the total number of direct
and indirect connections between all category pairs. We then used
the matrix of direct connections to build a map according to the
guidelines by [
23
,
27
] focusing on the most elicited connections.
We then used the indirect matrix to rene the map in case some
indirect connections with numerous occurrences did not appear in
the direct matrix. The resulting map is displayed as Figure 1 and
show the most important categories and connections.
3.3 Sample
After piloting and rening our interview protocol on 5 colleagues,
we recruited 22 participants through personal connections (for 2 of
them) and advertisement posters displayed throughout the campus
and in the city center. The mean age was 30.82 years old, with a
standard deviation of 11.29. In terms of educational level, one par-
ticipant had a French vocational professional degree. 50.09% of the
participants were Bachelor or Master students at an university. The
rest (36.36%) of the participants were at least university graduates.
In the population, we had 33% female participants. As for the coun-
try in which the participants grew up, 54.6% grew up in European
countries and 45.5% in Asian countries, thus showing high cultural
balance. 81% of the participants declared having at least some expe-
rience in the practice of music. Participants seemed pleased with
their participation; among 19 of them questioned, around 67% de-
clared that they learned new things about their musical tastes and
the underlying reasons for those during their interview.
4 RESULTS
Our interviews yielded 481 ladders: 62 (13%) terminated at the at-
tribute level, 241 (50%) at the consequence level, and 178 (37%) at
the personal value level. Thus, while people could not always relate
their music preferences up to the personal value level, multiple con-
crete connections between preferences and values could be found.
The Hierarchical Value Map, summarizing the main connections
that we found across the interviews, is given in Figure 1. It is to be
noted that besides those similarities, the interviews also showed
diversity among the respondents. Full information on connections
between pairs is given as Supplementary Material to this paper.
In existing music psychology literature, [
21
] argued that the
preferred music descriptor of people is genre. Indeed, the “Genre"
attribute occurred frequently in our interviews (used more than
200 times by more than 77% of the participants), and also has high
connectivity in the Hierarchical Value Map.
When coding our elements into overarching categories, we ini-
tially doubted whether the “Lyrics" and “Story" attributes should
be combined into a single linguistic category. However, when con-
sidering them as separate items, it can be noted they have clearly
dierent connections in the value map. Indeed, when reconsidering
the information expressed by the interviewees, “Lyrics" would con-
sider the poetic aspect of a song’s text, while “Story" rather would
consider whether a listener would feel being taken along with the
narrative of the music.
With respect to consequences, the most frequently mentioned
category was “Emotion(get)" (70 occurrences, mentioned by 95%
of the participants), representing any kind of emotions, mood or
feeling induced by the music. Thus, using music as an aective
503

RecSys ’19, September 16–20, 2019, Copenhagen, Denmark Manolios, Hanjalic & Liem, 2019
Figure 1: Hierarchical Value Map, based on our global interview data. Red links indicate a strong relation (cited 10+ times),
blue links a medium relation (cited between 9 and 5 times) and gray links a weak relation (cited 4 times).
inducer seems to be an important intermediate reason for listening
to certain music. However, as we can see on the map, further up,
this leads to dierent personal values, with Openness to Change
being the only clear direct link, while other links are indirect, via
other consequences. Frequently, “Emotion(get)" connected to “Re-
laxation", which both relates to the personal value of Conservation,
as well as the consequence of “Being Functional". Thus, relaxing is
both perceived as healthy (Conservation), and as a way to become
more ecient, ultimately connecting to Self-Enhancement.
Another interesting consequence is “Talent" which expresses
both the innate talent, but also the amount of eort artists put
in their performance or cultivating their skills. Its link to Self-
Transcendence and Art reveals that some people care about the
process, not just the nal product, and want to support and reward
artists who are working hard (as opposed to artists perceived to
create music that is easy to make and to sell).
Moving up to personal values, we observe that not all personal
values are equally well represented. Six personal values occurred
frequently enough (in at least 9 of the interviews) to appear in our
Hierarchical Value Map; out of these, Openness to Change was most
frequently mentioned (54 times by 15 (68%) of the participants). In
contrast, Hedonism was mentioned only 8 times by 6 (27%) of the
participants, and therefore does not appear in the map.
Moving back down from personal values to consequences and at-
tributes in our map, our results suggest that people who value
Openness to Change tend to enjoy more diverse and complex
music. Those whose musical taste is strongly connected to Self-
Transcendence would tend to care more about the message of the
song. Individuals caring about their Self-Enhancement tend to pre-
fer music that help them to achieve their objectives or improve
themselves (learn new things or resolve challenges), as well as mu-
sic that helps them to relax. Relaxing music might also be enjoyed
by people who care about Conservation, both regarding their own
safety and being a good member of the society. Social Connection
could be supported by music that is easily accessible (the ”Simple"
category) as well as music with a Meaning people can relate too.
Finally, people for whom Art is important by itself, would tend to
care about the amount of time and eort put into a song.
5 CONCLUSION AND FUTURE WORK
In this work, we presented a rst study into connecting music
preference to personal values, for potential future use in value-
based recommender system scenarios. As we showed, it indeed is
possible to establish connections between music preference and
personal values. Personal values are not the sole drivers of music
preference; 37% of our elicited ladders terminated at this level, while
50% of the ladders terminated at a consequence. At the same time,
most of our interviewees did express more than one personal value
during their interviews. As we expected, not all personal values
were equally inuential; from our interviews, 6 personal values
emerged as particularly dominant from our interviews. Openness
to Change especially turned out to be a frequently mentioned value;
this may be related to the academic skew in our sample.
A logical follow-up to our work is to assess the validity of emerg-
ing strong relations from our interviews, by conducting larger-scale
quantitative studies with more balanced user samples. For this, we
plan to establish datasets on actual music consumption and declared
preferences, and to relate these to important personal values.
For larger-scale quantitative studies, it will not be possible to
employ the laddering technique to connect preferences to values,
as this technique requires face-to-face interaction between two
humans. Therefore, other means will need to be found to elicit
personal values. For this, traditional survey instruments could be
used. However, our results suggest that only a subset of personal
values will need to be considered, so full survey instruments may
not be necessary, which is positive from a usability perspective. As
mentioned before, we ultimately strive to integrate our ndings
into novel value-based recommender systems, and to investigate
whether the value angle may lead to more challenging, yet accept-
able and useful personalized recommendations.
ACKNOWLEDGMENTS
We would like to thank all the persons who participated in our
study, as well as our colleague Francesca Lucas who helped us with
the coding of the results and our reviewers for their encouraging
and valuable comments.
504

The Influence of Personal Values on Music Taste
REFERENCES
[1]
Vimala Balakrishnan and Hossein Arabi. 2018. HyPeRM: A hybrid personality-
aware recommender for movie. Malaysian Journal of Computer Science 31, 1
(2018), 48–62.
[2]
Arielle Bonneville-Roussy, Peter J Rentfrow, Man K Xu, and Je Potter. 2013.
Music through the ages: Trends in musical engagement and preferences from ado-
lescence through middle adulthood. Journal of personality and social psychology
105, 4 (2013), 703.
[3]
Robert A Brown. 2012. Music preferences and personality among Japanese
university students. International Journal of Psychology 47, 4 (2012), 259–268.
[4]
Marc JMH Delsing, Tom FM Ter Bogt, Rutger CME Engels, and Wim HJ Meeus.
2008. Adolescents’ music preferences and personality characteristics. Euro-
pean Journal of Personality: Published for the European Association of Personality
Psychology 22, 2 (2008), 109–130.
[5]
Peter Gregory Dunn, Boris de Ruyter, and Don G Bouwhuis. 2012. Toward a
better understanding of the relation between music preference, listening behavior,
and personality. Psychology of Music 40, 4 (2012), 411–428.
[6]
Bruce Ferwerda, Emily Yang, Markus Schedl, and Marko Tkalcic. 2015. Personality
traits predict music taxonomy preferences. In Proceedings of the 33rd Annual ACM
Conference Extended Abstracts on Human Factors in Computing Systems. ACM,
2241–2246.
[7]
Darren George, Kelly Stickle, Faith Rachid, and Alayne Wopnford. 2007. The
association between types of music enjoyed and cognitive, behavioral, and per-
sonality factors of those who listen. Psychomusicology: A Journal of Research in
Music Cognition 19, 2 (2007), 32.
[8]
Fabian Greb, Wol Schlotz, and Jochen Steens. 2017. Personal and situational
inuences on the functions of music listening. Psychology of Music (2017),
0305735617724883.
[9]
Jonathan Gutman. 1982. A means-end chain model based on consumer catego-
rization processes. Journal of marketing 46, 2 (1982), 60–72.
[10]
Jill Harrison and John Ryan. 2010. Musical taste and ageing. Ageing & Society 30,
4 (2010), 649–669.
[11]
Rong Hu and Pearl Pu. 2011. Enhancing collaborative ltering systems with per-
sonality information. In Proceedings of the fth ACM conference on Recommender
systems. ACM, 197–204.
[12]
Mads Meier Jæger and Tally Katz-Gerro. 2008. The rise of the cultural omnivore
1964-2004. Research department of social policy and welfare services Working Paper
9 (2008).
[13]
Alexandra Langmeyer, Angelika Guglhör-Rudan, and Christian Tarnai. 2012.
What do music preferences reveal about personality? Journal of individual
dierences (2012).
[14]
Audrey Laplante. 2012. Who inuence the music tastes of adolescents?: a study
on interpersonal inuence in social networks. In Proceedings of the second in-
ternational ACM workshop on Music information retrieval with user-centered and
multimodal strategies. ACM, 37–42.
[15]
Audrey Laplante. 2014. Improving music recommender systems: What can we
learn from research on music tastes?. In ISMIR. 451–456.
[16]
Michael Leitner, Peter Wolkerstorfer, Reinhard Sefelin, and Manfred Tscheligi.
2008. Mobile multimedia: identifying user values using the means-end theory. In
Proceedings of the 10th international conference on Human computer interaction
with mobile devices and services. ACM, 167–175.
[17]
Adrian C North and David J Hargreaves. 1999. Music and adolescent identity.
Music education research 1, 1 (1999), 75–92.
[18]
Adrian C North and David J Hargreaves. 2007. Lifestyle correlates of musical
preference: 1. Relationships, living arrangements, beliefs, and crime. Psychology
of music 35, 1 (2007), 58–87.
[19]
Adrian C North and David J Hargreaves. 2007. Lifestyle correlates of musical
preference: 3. Travel, money, education, employment and health. Psychology of
music 35, 3 (2007), 473–497.
[20]
Ken Peers and Charles E Gengler. 2003. How to identify new high-payo
information systems for the organization. Commun. ACM 46, 1 (2003), 83–88.
[21]
Peter J Rentfrow and Samuel D Gosling. 2003. The do re mi’s of everyday life: the
structure and personality correlates of music preferences. Journal of personality
and social psychology 84, 6 (2003), 1236.
[22]
Peter J Rentfrow and Samuel D Gosling. 2007. The content and validity of music-
genre stereotypes among college students. Psychology of Music 35, 2 (2007),
306–326.
[23]
Thomas J Reynolds and Jonathan Gutman. 1988. Laddering theory, method,
analysis, and interpretation. Journal of advertising research 28, 1 (1988), 11–31.
[24]
Thomas J Reynolds and Joan M Phillips. 2009. A review and comparative analysis
of laddering research methods. In Review of Marketing Research. Emerald Group
Publishing Limited, 130–174.
[25]
Sonia Roccas, Lilach Sagiv, Shalom H Schwartz, and Ariel Knafo. 2002. The big
ve personality factors and personal values. Personality and social psychology
bulletin 28, 6 (2002), 789–801.
[26] Milton Rokeach. 1973. The nature of human values. Free press.
RecSys ’19, September 16–20, 2019, Copenhagen, Denmark
[27]
Abra Saaka, Chris Sidon, and Brian F Blake. 2004. Laddering. A "how to do it"
manual–With a note of caution. Research reports in consumer behavior: How to
series. Ohio: Cleveland State University (2004).
[28]
Thomas Schäfer and Claudia Mehlhorn. 2017. Can personality traits predict
musical style preferences? A meta-analysis. Personality and Individual Dierences
116 (2017), 265–273.
[29]
Shalom H Schwartz. 1992. Universals in the content and structure of values: The-
oretical advances and empirical tests in 20 countries. In Advances in experimental
social psychology. Vol. 25. Elsevier, 1–65.
[30]
Shalom H Schwartz. 2006. Basic human values: An overview. Recuperado de
http://www. yourmorals. org/schwartz (2006).
[31]
Marko Tkalcic and Li Chen. 2015. Personality and Recommender Systems. Springer
US, Boston, MA, 715–739. https://doi.org/10.1007/978-1- 4899-7637- 6_21
[32]
Marko Tkalcic, Matevz Kunaver, Andrej Košir, and Jurij Tasic. 2011. Addressing
the new user problem with a personality based user similarity measure. In First
International Workshop on Decision Making and Recommendation Acceptance
Issues in Recommender Systems (DEMRA 2011), Vol. 106.
[33]
Marko Tkalcic, Matevz Kunaver, Jurij Tasic, and Andrej Košir. 2009. Personal-
ity based user similarity measure for a collaborative recommender system. In
Proceedings of the 5th Workshop on Emotion in Human-Computer Interaction-Real
world challenges. 30–37.
[34]
Alexander Vander Stichele and Rudi Laermans. 2006. Cultural participation in
Flanders: Testing the cultural omnivore thesis with population data. Poetics 34, 1
(2006), 45–64.
[35]
Alan Warde, David Wright, and Modesto Gayo-Cal. 2007. Understanding cultural
omnivorousness: Or, the myth of the cultural omnivore. Cultural sociology 1, 2
(2007), 143–164.
[36]
Wen Wu, Li Chen, and Yu Zhao. 2018. Personalizing recommendation diversity
based on user personality. User Modeling and User-Adapted Interaction 28, 3 (2018),
237–276.
505