Conference PaperPDF Available

Towards Psychometrics-Based Friend Recommendations in Social Networking Services

Authors:
  • Telekom Innovation Laboratories / TU Berlin

Abstract

Two of the defining elements of Social Networking Services are the social profile, containing information about the user, and the social graph, containing information about the connections between users. Social Networking Services are used to connect to known people as well as to discover new contacts. Current friend recommendation mechanisms typically utilize the social graph. In this paper, we argue that psychometrics, the field of measuring personality traits, can help make meaningful friend recommendations based on an extended social profile containing collected smartphone sensor data. This will support the development of highly distributed Social Networking Services without central knowledge of the social graph.
Towards Psychometrics-based Friend
Recommendations in Social Networking Services
Felix Beierle, Kai Grunert, Sebastian G¨
ond¨
or, Viktor Schl¨
uter
Service-centric Networking
Telekom Innovation Laboratories / Technische Universit¨
at Berlin
Berlin, Germany
{beierle, kai.grunert, sebastian.goendoer}@tu-berlin.de
Technische Universit¨
at Berlin, Berlin, Germany
viktor.a.schlueter@campus.tu-berlin.de
Abstract—Two of the defining elements of Social Networking
Services are the social profile, containing information about the
user, and the social graph, containing information about the
connections between users. Social Networking Services are used
to connect to known people as well as to discover new contacts.
Current friend recommendation mechanisms typically utilize the
social graph. In this paper, we argue that psychometrics, the
field of measuring personality traits, can help make meaningful
friend recommendations based on an extended social profile
containing collected smartphone sensor data. This will support
the development of highly distributed Social Networking Services
without central knowledge of the social graph.
I. INTRODUCTION
Social Networking Services (SNSs) are one of the most
used services on the World Wide Web [1]. Two typical ele-
ments of a SNS are the social profile, containing information
about a user, for example her interests, and the social graph,
containing information about the connections between users.
In our previous work, we argued that the smartphone is the
optimal social networking device [2]. It typically has only
one user and, with recent developments in smartphone sensor
technologies and available APIs, more and more personal data
– like location traces, most frequently used apps, etc. – is
available that could potentially extend existing social profiles.
One of the typical applications in SNSs are friend rec-
ommendations. When recommending new connections in an
SNS, typically, the social graph is utilized [3]. While doing
so enables the incorporation of graph-based properties like the
number of mutual friends, there are also studies that look into
the similarity of attributes of neighboring nodes, thus incor-
porating the social profile in the recommendation process [4].
The basis for the cited studies about friend recommendations
is the insight that homophily – the tendency for people to
associate themselves with people who are similar to them –
is structuring any type of network [5]. Looking further into
the fields of psychology and social sciences, psychometrics,
the academic field that deals with measuring psychological
personality traits, seems like a promising research area provid-
ing results that could help improve friend recommendations
in SNSs. Recently, the company Cambridge Analytica was
in the media because of their alleged success in utilizing
psychometrics in targeted political campaign advertisements
[6], though their impact on the campaign remains somewhat
unclear [7]. Although the use case is different – targeted
advertising instead of friend recommendation – this shows
the potential of applying psychology research results to other
fields. In this paper, we argue that combining current smart-
phone technologies with findings from psychometrics will
enable meaningful friend recommendations based on social
profiles without requiring knowledge of the social graph. Our
main contributions of this short paper is a thorough analysis
of the theoretical background of psychometrics in relation to
SNSs and mobile devices, including a proposal of how to
integrate the insights into friend recommendations in SNSs.
II. ANALYS IS A ND CO NCEPT
In this section, we give a detailed analysis of relevant work
related to psychometrics, social networking, and smartphone
usage. In Section II-A, we give a literature review on how
and why people actually connect with each other in (offline)
social networks. We outline the concepts of homophily and
personality from psychology and social sciences. In order to
ensure that the same concepts hold true in SNSs, we look into
existing research on SNSs and personality in Section II-B. In
Section II-C, we investigate the relationship between smart-
phone usage and personality. As we will deal with attributes of
users rather then existing connections between users, we will
look at existing definitions and components of social profiles
in Section II-D.
A. Psychology and Social Sciences
In this section, we want to investigate two concepts from
psychology and social sciences: homophily and personality.
Those two concepts will help us conceptualize the parameters
we need in a system for psychometrics-based friend recom-
mendations. Furthermore, it will answer the questions ”When
do people become friends?”, i.e., ”When do people create
edges in a social graph?”, which are necessary to be asked
in social networking.
Homophily is the concept that people tend to associate
themselves with other people that are similar to them. Ac-
cording to McPherson et al., this principle structures network
ties of every type, including friendship, work, or partnership
[5]. Some of the categories in which people have homophilic
contacts are ethnicity, age, education, and gender.
The social profile is representing a user. The personality
of a person influences a multitude of aspects, e.g., job per-
formance, satisfaction, or romantic success [8] and is a ”key
determinant for the friendship formation process” [9]. One of
the established ways to talk and research about personality
is the so-called Big Five or Five-Factor model [10], [11].
The five personality factors spell the acronym OCEAN and
are openness to experience,conscientiousness,extraversion,
agreeableness, and neuroticism. In their study, Selfhout et al.
show the importance of homophily for friendship networks
[12]. For three of the five factors (openness to experience,
extraversion, and agreeableness), they conclude that people
tend to select friends with similar levels of those traits.
Additionally to friendship, there are several studies finding
correlations between different aspects of everyday life and the
five factors. Especially interesting for social networking related
questions is the correlations between the five factors and
preferences or interests. [13] and [14] are two of the studies
that find correlations between personality and the music the
persons prefer to listen to. As we will show in Section II-D,
music preference is a typical element for a social profile. It
is the most commonly filled attribute in publicly accessible
Facebook profiles after gender [15].
B. Social Networking Services and Personality
Several studies suggest that the findings about (offline)
social networks are also valid when dealing with SNSs. Liu
claims the social profile is a ”performance” by the user who
expresses herself by crafting the profile [16]. While this might
be true, various studies show that this does not imply that
this ”performance” distorts the personality that is expressed in
the profile. For example, Back et al. conclude in their study
that ”Facebook Profiles Reflect Actual Personality, Not Self-
Idealization,” as the title of their paper indicates [17]. In their
study, Goldbeck et al. show that Facebook profiles can be
used to predict personality [8]. Another study comes to the
same conclusion and shows ”that Facebook-based personality
impressions show some consensus for all Big Five dimen-
sions” [18].
C. Smartphone Usage and Personality
Some studies on pre-smartphone-era cell phones found cor-
relations between personality traits and mobile phone usage.
For example, these are the results of a study done on the
general use of mobile phones (calls, text messages, changing
ringtones and wallpapers) [19], as well as of a study about
using mobile phone games [20]. While these studies were
based on self reports by users, Chittaranjan et al. conducted
two user studies in which they collected usage data on Nokia
N95 phones [21], [22]. In those studies, the authors were
looking at Bluetooth scan data, call logs, text messages,
calling profiles, and application usage. At the time the study
was conducted, apps were not as common as nowadays with
Android and iOS. The authors state that ”features derived from
the App Logs were sparse due to the low frequency of usage
of some of the applications” [22]. It will be interesting to
compare the results from their study to a new study where
the usage of apps is commonplace. The results of the cited
studies indicate that ”several aggregated features obtained
from smartphone usage data can be indicators of the Big-Five
traits” [22].
In a more recent study, Lane and Manner showed relations
between the usage of apps and the five personality dimensions
[23]. Apps were categorized in different application types:
communication, games, multimedia, productivity, travel, and
utilities. Overall, the referenced studies indicate strong cor-
relations between smartphone usage behavior and personality
traits.
D. Social Profiles
The social profile is one of the central elements of SNSs.
In Boyd and Ellison’s definition of Social Network Sites, the
”public or semi-public profile within a bounded system” is
the first defining element, and the ”backbone” of the SNS
[24]. Typical elements of a social profile are ”age, location,
interests, and an ’about me’ section,” and a photo. In [25],
the social profile is the first defining functionality of an SNS.
Here, the authors call the functionality ”identity management,”
as the profile is a ”representation of the own person.” In a
survey paper about SNSs, the social profile is described as
the ”core” of an Online Social Network [26]. Another recent
survey describes the creation and maintenance of user profiles
as the ”basic functionality” of SNSs [27]. In [28], several
SNSs from different categories, like general (e.g. Facebook),
business oriented (e.g. LinkedIn) or special purpose (e.g.
Twitter), were analyzed. The social profile is an element that
is present in all of those SNSs. Rohani and Hock state that
the type of information included in social profiles differs
between different SNSs [29]. In their analysis of publicly
disclosed Facebook profile information, Farahbakhsh et al.
distinguish between personal and interest-based attributes [15].
Personal attributes include a friend list, current city, hometown,
gender, birthday, employers, college, and high school. Interest-
based attributes are music, movie, book, television, games,
teams, sports, athletes, activities, interests, and inspirations.
Lampe et al. distinguish between three different types of
information: referents, interests, and contact [30]. Referents
include verifiable attributes: hometown, high school, residence,
concentration. Contact information are also verifiable, for
example website, email, address, or birthday. As the authors
indicate, interests are less verifiable. Interests include an ’about
me’ section, favorite music, movies, TV shows, books, quotes,
and political views. As the cited studies in Section II-C and
also social networking related studies (e.g., [31]) suggest, more
detailed user data additional to the data typically available in
a social profile can help improve recommendations in SNSs.
III. CONCEPT AND PROTOTY PE
Research in psychology and social sciences indicates that
homophily in age, education, etc., as well as in personality
traits, is a strong indicator for friendship, i.e., for the creation
of an edge in a social graph. Several of the aforementioned
studies conclude with findings about correlations between
smartphone usage and personality traits. Combining those
insights, in order to make meaningful recommendations for
new connections in an SNS, we can recommend users that
show a similar behavioral pattern with their smarthphones. As
the existing studies suggest, this will indicate the similarity of
their personality traits. Doing this, we do not necessarily need
to know which behavior indicates which personality trait.
Such a mechanism for friend recommendations can have
several benefits: (1) By logging information about the smart-
phone usage behavior (a lot of which can be done unobtru-
sively, see [32]), we could automatically set up or update an
existing social profile in an SNS, eliminating or reducing the
tedious task to keep such a profile up to date [30]. (2) When
fully relying on social profile data for friend recommendations,
the social graph is not needed in the recommendation process.
This will enable the development of highly distributed SNSs,
for example in device-to-device scenarios where two randomly
meeting people could determine – without contacting some
centralized server – how similar they are and thus are recom-
mended to be friends. (3) By collecting highly personalized
user data, further studies in the fields of psychometrics could
be possible.
We are implementing an Android prototype. With the
Google Awareness API, Google offers to get the user’s current
time, location, nearby places, nearby beacons, headphone state,
activity, and weather.1Android also enables developers to
retrieve a list of the most frequently used apps. Most music
players broadcast what the user is listening to, so other
apps can retrieve this information. The available APIs and
mechanisms allow for an unobtrusive collection of user data
on Android smartphones. After collecting the mentioned data,
in order to estimate their similarity, two users can share their
data in a device-to-device manner by utilizing appropriate data
structures and technologies like Bluetooth, Wi-Fi Direct, or
Wi-Fi Aware.
IV. REL ATED WOR K
In this section, we review related work on smartphone
sensor data collection (Section IV-A) and on link prediction
in SNSs (Section IV-B).
A. Smartphone Data Collection
In [33], the authors survey the state of the art of ”Antic-
ipatory Mobile Computing,” describing how the advances in
mobile technology will enable predicting future contexts and
acting on it. This field has some similarities to our work,
especially with respect to collecting and using sensor data,
but it does not focus on social networking.
In [34], the authors present a framework called ”BaranC”
for monitoring and analyzing digital interaction of users with
their smartphones. The architecture uses cloud technologies
1https://developers.google.com/awareness/
to analyze data and thus raises privacy concerns. The goal
of BaranC is not social networking but offering personalized
services. In another work by the same authors, an application
utilizing their framework is presented [35], predicting the next
application a user will use.
Wang et al. present a system that collects a multitude of
sensor data from mobile devices and queries the users with
questionnaires. The collected data is then used to accurately
predict the GPA of the undergraduate students who partici-
pated [36]. Xiong et al. present a system for social sciences
studies that collects sensor data and enables researchers to
create surveys for study participants [37]. Again, in both
those cases, the developed concepts did not focus on social
networking.
B. Link Prediction
One of the common ways to research about links in social
networks is link prediction. The key difference to our work is
that here, the social graph is used to calculate the prediction or
recommendation, while our approach is also feasible in device-
to-device mobile SNS scenarios where the social graph is not
known. Yin et al. analyzed links in social networks based on
”intuition-based” aspects: homophily (shared attributes), rarity
(matching uncommon attributes), social influence (more likely
to link to person that shares attributes with existing friends),
common friendship (mutual friends), social closeness (being
close to each other in the social graph), preferential attachment
(more likely to link to a popular person) [3]. Most aspects
focus on the social graph or global knowledge about attribute
distribution (in the case of rarity). In the work by Mohajureen
et al., the authors use the attributes of neighboring nodes in
the friend recommendation process [4]. For this algorithm to
work, the social graph as well as the features of each user
have to be available.
A somewhat special case of link prediction or friend rec-
ommendation is described in WhozThat? [38]. Here, the idea
is to retrieve information about another person you just met.
Via Bluetooth, user handles from an SNS are exchanged and
data about the other person can be retrieved from that SNS.
As described in Section III, by following our concept, the
same scenario can be realized in a distributed manner without
contacting existing centralized SNS.
V. CONCLUSION AND FUTURE WORK
In this paper, we proposed the extension of social profiles
with smartphone sensor data. We showed that research results
from the field of psychometrics suggest that we then can
calculate relevant friend recommendations based on those
profiles, without utilizing the social graph. This will enable
recommendations in highly distributed SNSs.
Future work includes conducting a user study with our pro-
totype to confirm our conclusions. Furthermore, in the device-
to-device SNS scenario, the issue of privacy-awareness will
be further investigated so that similarity estimations between
users are possible without sharing all the collected sensitive
sensor data.
ACKNOWLEDGMENT
This work has received funding from the European Union’s
Horizon 2020 research and innovation program under grant
agreement No 645342, project reTHINK and from project
DYNAMIC2(grant No 01IS12056), which is funded as part
of the Software Campus initiative by the German Federal
Ministry of Education and Research (BMBF).
REFERENCES
[1] S. Greenwood, A. Perrin, and M. Duggan. (2016, Nov.) So-
cial Media Update 2016. http://www.pewinternet.org/2016/11/11/
social-media- update-2016/.
[2] F. Beierle, S. G¨
ond¨
or, and A. K¨
upper, “Towards a Three-tiered Social
Graph in Decentralized Online Social Networks,” in Proc. 7th Interna-
tional Workshop on Hot Topics in Planet-Scale mObile Computing and
Online Social neTworking (HotPOST). ACM, Jun. 2015, pp. 1–6.
[3] Z. Yin, M. Gupta, T. Weninger, and J. Han, “A Unified Framework for
Link Recommendation Using Random Walks,” in Proc. 2010 Interna-
tional Conference on Advances in Social Networks Analysis and Mining
(ASONAM). IEEE, Aug. 2010, pp. 152–159.
[4] M. Mohajireen, C. Ellepola, M. Perera, I. Kahanda, and U. Kanewala,
“Relational Similarity Model for Suggesting Friends in Online Social
Networks,” in Proc. 2011 6th International Conference on Industrial
and Information Systems (ICIIS). IEEE, Aug. 2011, pp. 334–339.
[5] M. McPherson, L. Smith-Lovin, and J. M. Cook, “Birds of a Feather:
Homophily in Social Networks,” Annual Review of Sociology, vol. 27,
pp. 415–444, 2001.
[6] R. Blakely. (2016, Sep.) Data scientists target 20 million
new voters for Trump. http://www.thetimes.co.uk/article/
trump-calls- in-brexit- experts-to- target-voters-pf0hwcts9.
[7] N. Confessore and D. Hakim. (2017, Mar.) Data Firm Says ‘Secret
Sauce’ Aided Trump; Many Scoff. https://www.nytimes.com/2017/03/
06/us/politics/cambridge-analytica.html.
[8] J. Golbeck, C. Robles, and K. Turner, “Predicting Personality with Social
Media,” in Proc. CHI ’11 Extended Abstracts on Human Factors in
Computing Systems (CHI EA). ACM, May 2011, pp. 253–262.
[9] S. Burgess, E. Sanderson, and M. Uma ˜
na-Aponte, School Ties: An
Analysis of Homophily in an Adolescent Friendship Network, ser. CMPO
Working Paper Series. Centre for Market and Public Organisation,
2011, no. 11/267.
[10] E. C. Tupes and R. E. Christal, “Recurrent Personality Factors Based
on Trait Ratings,Journal of Personality, vol. 60, no. 2, pp. 225–251,
Jun. 1992.
[11] R. R. McCrae and O. P. John, “An Introduction to the Five-Factor Model
and Its Applications,” Journal of Personality, vol. 60, no. 2, pp. 175–215,
1992.
[12] M. Selfhout, W. Burk, S. Branje, J. Denissen, M. Van Aken, and
W. Meeus, “Emerging Late Adolescent Friendship Networks and Big
Five Personality Traits: A Social Network Approach,Journal of Per-
sonality, vol. 78, no. 2, pp. 509–538, Apr. 2010.
[13] D. Rawlings and V. Ciancarelli, “Music Preference and the Five-Factor
Model of the NEO Personality Inventory,Psychology of Music, vol. 25,
no. 2, pp. 120–132, Oct. 1997.
[14] P. J. Rentfrow and S. D. Gosling, “The Do Re Mi’s of Everyday Life:
The Structure and Personality Correlates of Music Preferences.” Journal
of Personality and Social Psychology, vol. 84, no. 6, pp. 1236–1256,
2003.
[15] R. Farahbakhsh, X. Han, A. Cuevas, and N. Crespi, “Analysis of publicly
disclosed information in Facebook profiles,” in Proc. 2013 IEEE/ACM
International Conference on Advances in Social Networks Analysis and
Mining (ASONAM). ACM, Aug. 2013, pp. 699–705.
[16] H. Liu, “Social Network Profiles as Taste Performances,Journal of
Computer-Mediated Communication, vol. 13, no. 1, pp. 252–275, Oct.
2007.
[17] M. D. Back, J. M. Stopfer, S. Vazire, S. Gaddis, S. C. Schmukle,
B. Egloff, and S. D. Gosling, “Facebook Profiles Reflect Actual Per-
sonality, Not Self-Idealization,Psychological Science, vol. 21, no. 3,
pp. 372–374, Jan. 2010.
2http://www.dynamic-project.de
[18] S. D. Gosling, S. Gaddis, and S. Vazire, “Personality Impressions
Based on Facebook Profiles,” in Proc. International AAAI Conference
on Weblogs and Social Media (ICWSM). AAAI, 2007, pp. 1–4.
[19] S. Butt and J. G. Phillips, “Personality and self reported mobile phone
use,” Computers in Human Behavior, vol. 24, no. 2, pp. 346–360, Mar.
2008.
[20] J. G. Phillips, S. Butt, and A. Blaszczynski, “Personality and Self-
Reported Use of Mobile Phones for Games,” CyberPsychology &
Behavior, vol. 9, no. 6, pp. 753–758, Dec. 2006.
[21] G. Chittaranjan, J. Blom, and D. Gatica-Perez, “Who’s Who with Big-
Five: Analyzing and Classifying Personality Traits with Smartphones,
in Proc. 2011 15th Annual International Symposium on Wearable
Computers (ISWC). IEEE, Jun. 2011, pp. 29–36.
[22] ——, “Mining large-scale smartphone data for personality studies,”
Personal and Ubiquitous Computing, vol. 17, no. 3, pp. 433–450, Mar.
2013.
[23] W. Lane and C. Manner, “The influence of personality traits on mobile
phone application preferences,” Journal of Economics & Behavioral
Studies, vol. 4, no. 5, pp. 252–260, 2012.
[24] D. M. Boyd and N. B. Ellison, “Social Network Sites: Definition, His-
tory, and Scholarship,Journal of Computer-Mediated Communication,
vol. 13, no. 1, pp. 210–230, Oct. 2007.
[25] A. Richter and M. Koch, “Functions of Social Networking Services,
in Proc. 8th International Conference on the Design of Cooperative
Systems (COOP), May 2008, pp. 87–98.
[26] J. Heidemann, M. Klier, and F. Probst, “Online social networks: A survey
of a global phenomenon,” Computer Networks, vol. 56, no. 18, pp. 3866–
3878, Dec. 2012.
[27] T. Paul, A. Famulari, and T. Strufe, “A survey on decentralized Online
Social Networks,” Computer Networks, vol. 75, Part A, pp. 437–452,
Dec. 2014.
[28] S. G¨
ond¨
or, F. Beierle, S. Sharhan, H. Hebbo, E. Kucukbayraktar, and
A. K¨
upper, “SONIC: Bridging the Gap between Different Online Social
Network Platforms,” in Proc. 2015 IEEE International Conference on
Smart City/SocialCom/SustainCom (SmartCity). IEEE, Dec. 2015, pp.
399–406.
[29] V. A. Rohani and O. S. Hock, “On Social Network Web Sites: Defini-
tion, Features, Architectures and Analysis Tools,Journal of Computer
Engineering, vol. 1, pp. 3–11, 2009.
[30] C. A. Lampe, N. Ellison, and C. Steinfield, “A Familiar Face(Book):
Profile Elements As Signals in an Online Social Network,” in Proc.
SIGCHI Conference on Human Factors in Computing Systems (CHI).
ACM, Apr. 2007, pp. 435–444.
[31] J. Bao, Y. Zheng, D. Wilkie, and M. Mokbel, “Recommendations in
Location-based Social Networks: A Survey,Geoinformatica, vol. 19,
no. 3, pp. 525–565, 2015.
[32] F. Beierle, K. Grunert, S. G¨
ond¨
or, and A. K¨
upper, “Privacy-aware
Social Music Playlist Generation,” in Proc. 2016 IEEE International
Conference on Communications (ICC). IEEE, May 2016, pp. 5650–
5656.
[33] V. Pejovic and M. Musolesi, “Anticipatory Mobile Computing: A Survey
of the State of the Art and Research Challenges,” ACM Computing
Surveys, vol. 47, no. 3, pp. 1–29, Apr. 2015.
[34] M. Hashemi and J. Herbert, “User Interaction Monitoring and Analysis
Framework,” in Proc. 2016 IEEE/ACM International Conference on
Mobile Software Engineering and Systems (MobileSoft). ACM, 2016,
pp. 7–8.
[35] ——, “A Next Application Prediction Service Using the BaranC Frame-
work,” in Proc. 2016 IEEE 40th Annual Computer Software and
Applications Conference (COMPSAC). IEEE, Jun. 2016, pp. 519–523.
[36] R. Wang, G. Harari, P. Hao, X. Zhou, and A. T. Campbell, “SmartGPA:
How Smartphones Can Assess and Predict Academic Performance of
College Students,” in Proc. 2015 ACM International Joint Conference
on Pervasive and Ubiquitous Computing (UbiComp). ACM, Sep. 2015,
pp. 295–306.
[37] H. Xiong, Y. Huang, L. E. Barnes, and M. S. Gerber, “Sensus: A Cross-
platform, General-purpose System for Mobile Crowdsensing in Human-
subject Studies,” in Proc. 2016 ACM International Joint Conference on
Pervasive and Ubiquitous Computing (UbiComp). ACM, Sep. 2016,
pp. 415–426.
[38] A. Beach, M. Gartrell, S. Akkala, J. Elston, J. Kelley, K. Nishimoto,
B. Ray, S. Razgulin, K. Sundaresan, B. Surendar, M. Terada, and R. Han,
“WhozThat? Evolving an Ecosystem for Context-Aware Mobile Social
Networks,” IEEE Network, vol. 22, no. 4, pp. 50–55, 2008.
... In the field of psychology, the principle of homophily states that individuals tend to bond if they are similar. This tendency also holds true within social networks services employing digital communication [3]. ...
... Nowadays, a large portion of digital communication happens on smartphones which harbor highly personalized sensor and personal context data such as visited locations, music listened to, or app usage statistics [4]. Profiling users upon this data allows to estimate user similarities [3], [5]. A very commonly used -yet virtually never mined -resource on smartphones for user profiling is text messaging 1 or simply texting. ...
... In the field of psychology, the principle of homophily states that individuals tend to bond if they are similar. This tendency also holds true within social networks services employing digital communication [3]. ...
... Nowadays, a large portion of digital communication happens on smartphones which harbor highly personalized sensor and personal context data such as visited locations, music listened to, or app usage statistics [4]. Profiling users upon this data allows to estimate user similarities [3], [5]. A very commonly used -yet virtually never mined -resource on smartphones for user profiling is text messaging 1 or simply texting. ...
Preprint
Full-text available
In the field of social networking services, finding similar users based on profile data is common practice. Smartphones harbor sensor and personal context data that can be used for user profiling. Yet, one vast source of personal data, that is text messaging data, has hardly been studied for user profiling. We see three reasons for this: First, private text messaging data is not shared due to their intimate character. Second, the definition of an appropriate privacy-preserving similarity measure is non-trivial. Third, assessing the quality of a similarity measure on text messaging data representing a potentially infinite set of topics is non-trivial. In order to overcome these obstacles we propose affinity, a system that assesses the similarity between text messaging histories of users reliably and efficiently in a privacy-preserving manner. Private texting data stays on user devices and data for comparison is compared in a latent format that neither allows to reconstruct the comparison words nor any original private plain text. We evaluate our approach by calculating similarities between Twitter histories of 60 US senators. The resulting similarity network reaches an average 85.0% accuracy on a political party classification task.
... E-Smalltalker aims at incentivizing real-life smalltalk by exchanging interests profiles between users in proximity [15]. In our own previous work, we proposed using all available smartphone data and automatically comparing it in order to determine similarity in terms of interest and personality [1], [16]. In [17], Yang and Hwang proposed a mobile recommender system for point-of-interest (POI) recommendations that utilizes data exchanged in a device-to-device manner. ...
Article
Full-text available
Recommender systems recommend new movies, music, restaurants, etc. Typically, service providers organize such systems in a centralized way, holding all the data. Biases in the recommender systems are not transparent to the user and lock-in effects might make it inconvenient for the user to switch providers. In this paper, we present the concept, design, and implementation of MobRec, a mobile platform that decentralizes the data collection, data storage, and recommendation process. MobRec's architecture does not need any backend and solely consists of the users' smartphones, which already contain the users' preferences and ratings. Being in proximity in public places or public transportation, data is exchanged in a device-to-device manner, building local databases that can recommend new items. One of biggest challenges of such a system is the implementation of unobtrusive device-to-device data exchange on off-the-shelf Android devices and iPhones. MobRec facilitates such data exchange, building on Google Nearby Messages with Bluetooth Low Energy. We achieve the successful exchange of data within 3 to 4 minutes, making it suitable for the described scenario. We demonstrate the feasibility of decentralized recommender systems and provide blueprints for the development of seamless multi-platform device-to-device communication.
... In our own previous work, we outlined a general concept for a social networking service utilizing context data from smartphones, focusing on how connections between users can be established [29]. In [30], we proposed to utilize the relationship between smartphone data and personality traits for friend recommendations in device-to-device scenarios. ...
... Psychological research suggests that there are links between personality traits and everyday preferences (Beierle et al. 2017). With a smartphone, we will be able to track different types of data that might reflect the user's personality: the smartphone's sensors can track the user's physical context and the operating system can track the user's interaction with the smartphone and its apps. ...
Article
Full-text available
Context-aware applications stemming from diverse fields like mobile health, recommender systems, and mobile commerce potentially benefit from knowing aspects of the user’s personality. As filling out personality questionnaires is tedious, we propose the prediction of the user’s personality from smartphone sensor and usage data. In order to collect data for researching the relationship between smartphone data and personality, we developed the Android app track your daily routine (TYDR), which tracks and records smartphone data and utilizes psychometric personality questionnaires. With TYDR, we track a larger variety of smartphone data than many other existing apps, including metadata on notifications, photos taken, and music played back by the user. Based on the development of TYDR, we introduce a general context data model consisting of four categories that focus on the user’s different types of interactions with the smartphone: physical conditions and activity, device status and usage, core functions usage, and app usage. On top of this, we developed the Privacy Model for Mobile Data Collection Applications (PM-MoDaC) specifically tailored for apps that are related to the collection of mobile data, consisting of nine proposed privacy measures. We present the implementation of all of those measures in TYDR. Our experimental evaluation is based on data collected with TYDR during a two-month period. We find evidence that our users accept our proposed privacy model. Based on data about granting TYDR all or no Android system permissions, we find evidence that younger users tend to be less willing to share their data (average age of 30 years compared to 35 years). We also observe that female users tend to be less willing to share data compared to male users. We did not find any evidence that education or personality traits are a factor related to data sharing. TYDR users score higher on the personality trait openness to experience than the average of the population, which we assume to be evidence that the type of app influences the user base it attracts in terms of average personality traits.
... In our own previous work, we outlined a general concept for a social networking service utilizing context data from smartphones, focusing on how connections between users can be established [29]. In [30], we proposed to utilize the relationship between smartphone data and personality traits for friend recommendations in device-to-device scenarios. ...
Preprint
Full-text available
Typically, recommender systems from any domain, be it movies, music, restaurants, etc., are organized in a centralized fashion. The service provider holds all the data, biases in the recommender algorithms are not transparent to the user, and the service providers often create lock-in effects making it inconvenient for the user to switch providers. In this paper, we argue that the user's smartphone already holds a lot of the data that feeds into typical recommender systems for movies, music, or POIs. With the ubiquity of the smartphone and other users in proximity in public places or public transportation, data can be exchanged directly between users in a device-to-device manner. This way, each smartphone can build its own database and calculate its own recommendations. One of the benefits of such a system is that it is not restricted to recommendations for just one user - ad-hoc group recommendations are also possible. While the infrastructure for such a platform already exists - the smartphones already in the palms of the users - there are challenges both with respect to the mobile recommender system platform as well as to its recommender algorithms. In this paper, we present a mobile architecture for the described system - consisting of data collection, data exchange, and recommender system - and highlight its challenges and opportunities.
... The 2016 campaigns of Ted Cruz and Donald Trump employed CA who were also connected to the Leave campaign in the UK "Brexit" referendum [32]. Beierle et al. [6] point to the widespread belief that CA played a significant role in the election of Donald Trump using the psychometrics model. ...
Article
Full-text available
Targeted social media advertising based on psychometric user profiling has emerged as an effective way of reaching individuals who are predisposed to accept and be persuaded by the advertising message. This article argues that in the case of political advertising, this may present a democratic and ethical challenge. Hypertargeting methods such as psychometrics can “crowd out” political communication with opposing views due to individual attention and time limitations, creating inequities in the access to information essential for voting decisions. The use of psychometrics also appears to have been used to spread both information and misinformation through social media in recent elections in the U.S. and Europe. This article is an applied ethics study of these methods in the context of democratic processes and compared to purely commercial situations. The ethical approach is based on the theoretical, contractarian work of John Rawls, which serves as a lens through which the author examines whether the rights of individuals, as Rawls attributes them, are violated by this practice. The article concludes that within a Rawlsian framework, use of psychometrics in commercial advertising on social media platforms, though not immune to criticism, is not necessarily unethical. In a democracy, however, the individual cannot abandon the consumption of political information, and since using psychometrics in political campaigning makes access to such information unequal, it violates Rawlsian ethics and should be regulated.
... Psychological research suggests that there are links between personality traits and everyday preferences [4]. With the smartphone, we will be able to track different types of data that might reflect the user's personality: the smartphone's sensors can track the user's physical context and the operating system can track the user's interaction with the smartphone and its apps. ...
Article
Full-text available
Context-aware applications stemming from diverse fields like mobile health, recommender systems, and mobile commerce potentially benefit from knowing aspects of the user’s personality. As filling out personality questionnaires is tedious, we propose the prediction of the user’s personality from smartphone sensor and usage data. In order to collect data for researching the relationship between smartphone data and personality, we developed the Android app TYDR (Track Your Daily Routine) which tracks smart-phone data and utilizes psychometric personality questionnaires. With TYDR, we track a larger variety of smartphone data than similar existing apps, including metadata on notifications, photos taken, and music played back by the user. For the development of TYDR, we introduce a general context data model consisting of four categories that focus on the user’s different types of interactions with the smartphone: physical conditions and activity, device status and usage, core functions usage, and app usage. On top of this, we develop the privacy model PM-MoDaC specifically for apps related to the collection of mobile data, consisting of nine proposed privacy measures. We present the implementation of all of those measures in TYDR. Although the utilization of the user’s personality based on the usage of his or her smartphone is a challenging endeavor, it seems to be a promising approach for various types of context-aware mobile applications.
Article
Due to the dramatic rise in the popularity of the social media, the problem of analyzing the interconnections of digital traces and the psychological characteristics of the user become urgent. The development of deep learning models has ensured the automatic image analysis and feature extraction. In this study, we propose a clustering model for digital traces of users based on VGG16 deep learning model and a K-means clustering algorithm. We use this model for predictive analytics of digital footprints in the form of arbitrary images. The scientific novelty of the research lies in the originality of the proposed methodology. This technique firstly consists in directly using the compressed representation obtained by the encoder instead of object tags. And secondly, we propose a method for obtaining representations of clusters, based on the weighting of users belonging to a given cluster. To study the correlations of the user’s personality characteristics and their posted images, dataset of BIG Five traits of VK social network users was used. According to the results of the expert assessment of a specialist psychologist, the presented relations between the clusters content and the personality characteristics of the users were determined. Namely, the “faces” cluster is characterized by a high value of consciousness, the “cats” cluster is characteristic of introverted users, and low agreeableness was associated with the “landscapes” cluster. Based on the obtained similarity of cluster representations, the personality traits of users related to these clusters were found similar, and the possibility to combine these clusters was proposed.
Conference Paper
Thanks to the growing popularity and functionality, smartphone has become rapidly valuable potential tool for human behavior research, e.g., friendship relationship recognition, friend ship prediction, etc. Until recently, there has been many research efforts to study this issue using the sensed data collected from smartphones. However, almost previous works in finding friendship strength are based on several physical features or a few dimensions, such as using Bluetooth scanning and demographic data to explain friendship. Actually, friendship is complicated and coupled with many factors, such as physical propinquity, social, physical and psychological homophily. So, it is necessary and beneficial to examine it comprehensively, by taking into account all the involved factors. Aiming at closing part of this research gap, in this paper, from cross-space perspectives, we launch a friendship relationship study with smartphonebased sensing paradigm from cyber space, physical mobility, and personality trait homophily. By integrating the involved heterogeneous interactions, we propose a Deep AutoEncoderbased unified framework to predict the strength of friendship connections between users, where the friendship strength is categorized and asymmetrical. We conduct extensive experiments on a practically collected sensing data set, and show the efficiency and effectiveness of our proposed approaches.
Conference Paper
Full-text available
An overview is given of a user interaction monitoring and analysis framework called BaranC. Monitoring and analysing human-digital interaction is an essential part of developing a user model as the basis for investigating user experience. The primary human-digital interaction, such as on a laptop or smartphone, is best understood and modelled in the wider context of the user and their environment. The BaranC framework provides monitoring and analysis capabilities that not only records all user interaction with a digital device (e.g. smartphone), but also collects all available context data (such as from sensors in the digital device itself, a fitness band or a smart appliances). The data collected by BaranC is recorded as a User Digital Imprint (UDI) which is, in effect, the user model and provides the basis for data analysis. BaranC provides functionality that is useful for user experience studies, user interface design evaluation, and providing user assistance services. An important concern for personal data is privacy, and the framework gives the user full control over the monitoring, storing and sharing of their data.
Conference Paper
Full-text available
The burden of entry into mobile crowdsensing (MCS) is prohibitively high for human-subject researchers who lack a deeply technical orientation. As a result, the benefits of MCS remain beyond the reach of research communities (e.g., psychologists) whose expertise in the study of human behavior might advance application and understanding of MCS systems. This paper presents Sensus, a new MCS system for human-subject studies that bridges the gap between human-subject researchers and MCS methods. Sensus alleviates technical burdens with on-device, GUI-based design of sensing plans, simple and efficient distribution of sensing plans to study participants, and maximally uniform participant experience across iOS and Android devices. Sensing plans support many hardware and software sensors, automatic deployment of sensor-triggered surveys, and double-blind assignment of participants within randomized controlled trials. Sensus offers these features to study designers without requiring knowledge of markup and programming languages. We demonstrate the feasibility of using Sensus within two human-subject studies, one in psychology and one in engineering. Feedback from non-technical users indicates that Sensus is an effective and low-burden system for MCS-based data collection.
Conference Paper
Full-text available
Two of the most popular applications of smartphones are online social networking and playing back music. In this paper, we present the design and implementation of a prototype that combines those usages by creating an architecture that allows the generation and playback of group music playlists that are based on the musical taste of individual guests attending a meeting. In our architecture, we utilize automatically collected data on smartphones for the automatized generation of group music playlists. We follow the idea of utilizing context data in a preprocessing step to generate a group music profile for the recommendation process that generates a group music playlist. For designing such an architecture, we consider current discussions on privacy, data ownership, and data control.
Conference Paper
Full-text available
Online Social Network (OSN) platforms have become an important part of our everyday online lives. We communicate, share content, and organize meetings and events using social platforms and services. However, even though there is a strong trend towards OSN services to become the main communication medium, most OSN platforms are still proprietary, closed services that keep users from connecting directly and seamlessly to the services of other OSN platforms. The resulting lock-in effects are intentionally created by OSN operators, as their business models are built mostly on targeted advertisement services. While open and decentralized communication protocols exist for most other aspects of digital social interaction, there are only few micro-standards and protocols in existence for the social web. A holistic standard is yet missing. We envision a truly open and decentralized ecosystem of OSN platforms, where users are not cut off from friends using other social platforms and can freely migrate from one OSN platform to another at any time without losing established relationships in the social graph. This would allow users to freely choose an OSN platform of their liking instead of being limited in their choice to the platform used by one's friends. In this paper, we give an overview about the SOcial Network InterConnect protocol (SONIC), a novel social protocol for seamless social cross-platform communication. SONIC proposes a decentralized and heterogeneous Online Social Network Federation (OSNF), in which platforms are connected via a communication protocol. This allows different OSN platforms to seamlessly communicate with each other, while giving users the ability to migrate social profiles between platforms on demand without losing previously established connections to other user profiles.
Conference Paper
Full-text available
Online Social Networks have become one of the main tools for interpersonal online communication. In the age of the smartphone, mobile user scenarios become more and more important for Online Social Networks. Smartphones enable location-based and context-aware services, but bring the increased risk of privacy violations - at least in centralized OSN architectures. Decentralized Online Social Network architectures are promising as they inherently offer better privacy and less dependence on a single service provider, but they bring new challenges regarding core features of Online Social Networks. In this paper, we introduce a three-tiered view of the social graph and propose a new architecture for decentralized Online Social Networking applications, supporting the three-tiered view and focusing on location-based and context-aware user scenarios.
Article
Full-text available
Recent advances in localization techniques have fundamentally enhanced social networking services, allowing users to share their locations and location-related contents, such as geo-tagged photos and notes. We refer to these social networks as location-based social networks (LBSNs). Location data bridges the gap between the physical and digital worlds and enables a deeper understanding of users’ preferences and behavior. This addition of vast geo-spatial datasets has stimulated research into novel recommender systems that seek to facilitate users’ travels and social interactions. In this paper, we offer a systematic review of this research, summarizing the contributions of individual efforts and exploring their relations. We discuss the new properties and challenges that location brings to recommender systems for LBSNs. We present a comprehensive survey analyzing 1) the data source used, 2) the methodology employed to generate a recommendation, and 3) the objective of the recommendation. We propose three taxonomies that partition the recommender systems according to the properties listed above. First, we categorize the recommender systems by the objective of the recommendation, which can include locations, users, activities, or social media. Second, we categorize the recommender systems by the methodologies employed, including content-based, link analysis-based, and collaborative filtering-based methodologies. Third, we categorize the systems by the data sources used, including user profiles, user online histories, and user location histories. For each category, we summarize the goals and contributions of each system and highlight the representative research effort. Further, we provide comparative analysis of the recommender systems within each category. Finally, we discuss the available data-sets and the popular methods used to evaluate the performance of recommender systems. Finally, we point out promising research topics for future work. This article presents a panorama of the recommender systems in location-based social networks with a balanced depth, facilitating research into this important research theme.
Article
Full-text available
Understanding mobile phone users' preferences and behavior is essential for the commercial success of new application development. This study aims to enhance this understanding by identifying the personality traits associated with smart phone application use. Multiple regressions were used to analyze results from a sample of 233 participants. Consistent with recent personality research, we found that the "Big Five" personality dimensions are related to the application of smartphone technology. Extroverted individuals reported greater importance on gaming applications, but they viewed productivity applications as less important. Also, neurotics placed greater importance on travel applications, while less conscientious people indicated that communication, productivity, and utilities applications were less important to them.
Article
The current study focuses on the emergence of friendship networks among just-acquainted individuals, investigating the effects of Big Five personality traits on friendship selection processes. Sociometric nominations and self-ratings on personality traits were gathered from 205 late adolescents (mean age=19 years) at 5 time points during the first year of university. SIENA, a novel multilevel statistical procedure for social network analysis, was used to examine effects of Big Five traits on friendship selection. Results indicated that friendship networks between just-acquainted individuals became increasingly more cohesive within the first 3 months and then stabilized. Whereas individuals high on Extraversion tended to select more friends than those low on this trait, individuals high on Agreeableness tended to be selected more as friends. In addition, individuals tended to select friends with similar levels of Agreeableness, Extraversion, and Openness.
Conference Paper
Many cognitive, behavioral, and environmental factors impact student learning during college. The SmartGPA study uses passive sensing data and self-reports from students’ smartphones to understand individual behavioral differences between high and low performers during a single 10-week term. We propose new methods for better understanding study (e.g., study duration) and social (e.g., partying) behavior of a group of undergraduates. We show that there are a number of important behavioral factors automatically inferred from smartphones that significantly correlate with term and cumulative GPA, including time series analysis of activity, conversational interaction, mobility, class attendance, studying, and partying. We propose a simple model based on linear regression with lasso regularization that can accurately predict cumulative GPA. The predicted GPA strongly correlates with the ground truth from students’ transcripts (r = 0.81 and p < 0.001) and predicts GPA within ±0.179 of the reported grades. Our results open the way for novel interventions to improve academic performance.
Article
Abstract Because of growing popularity of Online Social Networks (OSNs) and huge amount of sensitive shared data, preserving privacy is becoming a major issue for OSN users. While most OSNs rely on a centralized architecture, with an omnipotent Service Provider, several decentralized architectures have recently been proposed for decentralized OSNs (DOSNs). In this work, we present a survey of existing proposals. We propose a classification of previous work under two dimensions: (i) types of approaches with respect to resource provisioning devices and (ii) adopted strategies for three main technical issues for DOSN (decentralizing storage of content, access control and interaction/signaling). We point out advantages and limitations of each approach and conclude with a discussion on the impact of DOSNs on users, OSN providers and other stakeholders.