Conference PaperPDF Available

Towards a Three-tiered Social Graph in Decentralized Online Social Networks

Authors:

Abstract

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.
Towards a Three-tiered Social Graph in
Decentralized Online Social Networks
Felix Beierle
Service-centric Networking
Technische Universität Berlin
Telekom Innovation
Laboratories
Berlin, Germany
beierle@tu-berlin.de
Sebastian Göndör
Service-centric Networking
Technische Universität Berlin
Telekom Innovation
Laboratories
Berlin, Germany
sebastian.goendoer@tu-
berlin.de
Axel Küpper
Service-centric Networking
Technische Universität Berlin
Telekom Innovation
Laboratories
Berlin, Germany
axel.kuepper@tu-
berlin.de
ABSTRACT
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 in-
creased 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, sup-
porting the three-tiered view and focusing on location-based
and context-aware user scenarios.
Categories and Subject Descriptors
C.2.4 [Computer-Communication Networks]: Dis-
tributed Systems; H.3.5 [Information Storage and Re-
trieval]: Online Information Services
General Terms
Design
Keywords
online social networks; mobile social networks; social graph;
location-based services; context-aware services
1. INTRODUCTION
Current Online Social Network s (OSN) utilize centralized
architectures and store any and all information about their
Permission to make digital or hard copies of all or part of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for profit or commercial advantage and that copies bear this notice and the full cita-
tion on the first page. Copyrights for components of this work owned by others than
ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or re-
publish, to post on servers or to redistribute to lists, requires prior specific permission
and/or a fee. Request permissions from permissions@acm.org.
HotPOST’15, June 22, 2015, Hangzhou, China.
Copyright c
2015 ACM 978-1-4503-3517-1/15/06 ...$15.00.
DOI: http://dx.doi.org/10.1145/2757513.2757517.
users. The resulting loss of control over one’s personal infor-
mation and the resulting lack of privacy is a major concern
for many users. In centralized OSN architectures, lack of
privacy is often an issue, because the operator of the OSN
might base their business model on the users’ private data,
while the users might not have intended to share their data
to that extent [8, 6, 4]. Furthermore, the usage of propri-
etary interfaces prohibits communication between different
OSN service providers, creating a well calculated lock-in ef-
fect for their users. Moreover, with the steep rise in the
processing power and enhanced features of smartphones as
well as their fast adoption by users, OSNs have moved mas-
sively to the mobile world (inspiring the term Mobile Social
Networks (MSN) for OSNs focused on mobile scenarios [11,
3]). This has opened new opportunities for OSNs and their
users, but also brings increased privacy risks. For instance,
many location-based and context-aware user scenarios are
only possible in MSNs, but the amount of sensitive data col-
lected, stored, and shared typically increases in these sce-
narios.
In this paper, we address these aspects of OSNs. In par-
ticular, we argue that distributing the social graph, which is
one of the core pieces of information of any social network, is
a vital step towards better privacy. We also argue that, ad-
ditionally, the capabilities of smartphones can be leveraged
to achieve both more flexible use of context information and
higher privacy of sensitive data for users of OSNs.
We investigate the relevant core features of OSNs and re-
view existing architectural approaches for implementing on-
line social networks. Based on this, we propose a three-tiered
view of the social graph and a new architectural concept,
called T3 architecture, supporting this three-tiered view. For
offering location-based and context-aware services and tak-
ing care of better privacy protection at the same time, our
approach is to use a global directory for retrieving informa-
tion on how to contact other users, and to use additional
decentralized directories that can offer search and recom-
mendations for specific contexts or locations. Thus, in sum-
mary, the three main contributions of this paper are (1)
the elaboration of core features of OSNs, (2) the proposal
to view the social graph as a three-tiered concept, and (3)
an architectural approach for distributed online social net-
work applications supporting the three-tiered social graph
and location-based and context-aware user scenarios.
The rest of this paper is structured as follows: After elab-
orating the core features of OSNs (Section 2) and investigat-
ing existing architectures for OSNs (Section 3), we develop
our three-tier view of the social graph in Section 4 and in-
troduce the T3 architecture in Section 5. In Section 6, we
conclude and point out future work.
2. FROM ONLINE TO MOBILE SOCIAL
NETWORKS
After analyzing user scenarios for OSNs for smartphone
users and existing definitions for OSNs and MSNs, we define
and discuss the core features of OSNs.
2.1 Mobile OSN User Scenarios
Before detailing the scenarios, we first explain why
the smartphone is the essential social networking device:
McLuhan imagined each technology as an extension of man
[13]: if the car is the extension of feet, glasses an exten-
sion of the eyes, then the smartphone is most certainly an
extension of all parts that constitute a person’s social inter-
actions. We believe that while the smartphone is capable of
being such an extension of man, current OSN architectures
or applications do not reflect this sentiment. Carrying the
smartphone at all times - with its sensors, processing power,
and connectivity like Bluetooth and WiFi-Direct - especially
local services do not need to rely on cloud services or tradi-
tional OSNs.
Imagine Alice, a smartphone user living in a larger city.
Going through her daily life, Alice is collecting data on her
smartphone, e.g., visited locations or artists she listened to.
By collecting this data, Alice’s interests are expressed im-
plicitly rather than explicitly as it is done by clicking a ”like”-
button. Going shopping, she can share selected parts of her
collected context and location data to receive personalized
recommendations. After shopping, imagine Alice receiving a
notification from Bob, who wants to play baseball. She does
not play baseball herself, but her smartphone can calculate
and suggest friends of hers that like baseball, and are not
friends with Bob. Alice can decide to forward recommended
missing players.
Studying at a university, Alice can talk to the people that
are in the same class. Going to the library to study, she can
look for people who are both at the library and in her classes.
Going downtown in the evening, when visiting a crowded
place, she can chat with strangers that are in her proximity,
e.g., to figure out what is at the center of the crowd. At
night, Alice goes to a party and quickly realizes that she
does not know many people there. She can retrieve infor-
mation about the people around her and find out mutual
contacts. Traveling to another city, Alice should be able to
query a public computer at her travel destination for recom-
mendations of local attractions or restaurants. Furthermore,
current contact information, like the postal address, could
be maintained locally with the users, making an up-to-date
version available for contacts.
2.2 Definitions of OSNs and MSNs
Some related work distinguishes between OSN and MSN.
In this section, we analyze the different definitions of the
two terms.
In their paper about decentralized OSNs, Paul et al. de-
fine OSNs as providing a form of profile management,re-
lationship management between users, interaction between
users, as well having an API (for interaction with third party
applications), search,recommendations, and social network
connector s (for connections between different OSNs) [14]. In
contrast to [14], we argue that an API is not essential for the
user of an OSN and is thus not necessarily a defining fea-
ture. Social network connectors contradict the lock-in effect
that centralized OSN platforms desire; for an approach to-
wards interconnection between OSNs, see [9]. Furthermore,
the authors of [14] do not consider the importance of mobile
user scenarios as described in Section 2.1.
In [11], Kayastha et al. define Mobile Social Networks as a
combination of OSNs with location-based services and con-
nections via Bluetooth and Wi-Fi. In [3], Bellavista et al.
define MSNs as OSNs that offer the possibility for oppor-
tunistic communities based on location information, with
the possibility of not needing a connection to a server. On
a less technical level, they define an MSN as the combina-
tion of virtual communities (”virtual social spaces”) of OSNs
with ”physical social spaces.”
Aside from architectural implications regarding the con-
nections between mobile devices, we consider context and
location information to be the essential new element that
mobile usage and smartphones bring into the world of OSNs.
We argue that with the pervasiveness of smartphones, the
mobile aspect of an OSN is indispensable, even for tradi-
tional OSNs like Facebook. In [11], Facebook, typically re-
garded as an OSN, is defined as an MSN, as it offers location-
based services, and is reachable via a mobile website or mo-
bile application. We therefore use the terms OSN and MSN
synonymously.
2.3 Core Features
Motivated by the analysis and observations above, we dis-
tinguish between the following six core features for OSNs:
Personal Profile. Each user has a personal profile in
which he/she can describe oneself, typically with a pro-
file picture.
Social Graph. The social graph consists of the connections
between a user of the OSN with other users or prod-
ucts.
Interaction. Messages, updates, posts, or likes by friends
are typically displayed to linked users.
Context Information. The user’s context, e.g., current
location, availability for instant messaging, etc., is of-
ten displayed for linked users or used for group cre-
ation.
Search. Search for specific users or users with a similar
context.
Recommendations. Recommendations of friends or prod-
ucts, based on social graph and context information, is
the most common feature with which users can explore
the network around them.
There are different OSNs with different focuses.
LinkedIn1, for example, focuses on work relationships and
career opportunities, Twitter2is about - typically publicly
- broadcasting short messages. For each OSN, the six listed
1https://www.linkedin.com/
2https://twitter.com/
features might be of differing importance, and might have
differing complexity.
With the rise of smartphone devices in recent years, In-
stant Messenger applications seem to fulfill the role of OSNs,
with messaging apps like WeChat3or WhatsApp offering
some OSN features like a short profile. Looking closer, the
contact list or address book of each of those applications -
or the phone’s contact list with PTSN telephone numbers -
is actually a part of a social graph: Each user sees just di-
rect connections, he does not see people more than one hop
away. A social graph that is stored in a distributed way of-
fers inherently more privacy to the users of the OSN. At the
same time, recommendations that are based on knowledge
about the graph, pose a new challenge.
Today, each OSN or messaging application is vertically in-
tegrated: They have their own identity management, their
own social graph, and - most often - proprietary communi-
cation protocols. Thus, each OSN acts like a silo, with users
being bound to restrictions their OSN of choice comes with.
The modern smartphone is an optimal online social net-
working device. It contains a camera, sensors that register
the user’s context, it can connect to the Internet, and it can
display websites, videos, or images that other users posted.
The generated context information especially deserves pro-
tection, as it can contain the most private information like
health data or exact user locations at any time. Given the
processing power of current smartphones, they can be used
to locally process data, e.g., lots of the context data a phone
generates. For instance, a phone could track and rank vis-
ited places or favorite artists, independent of specific OSN
applications, leaving the user in control of his data, letting
him decide when to share what data. Additionally, collected
context and location data offers the possibility to create dy-
namic social graphs, connecting users without them having
to explicitly befriend each other. This way, people who are
in the same city and are interested in a specific style of music
genre could automatically be connected and share informa-
tion about local shows and concerts.
3. ARCHITECTURAL APPROACHES
In this section, we motivate the utilization of decentralized
system architectures, before we summarize current research
about OSNs and MSNs.
3.1 On Decentralized Architectures
In the beginning of the 1990s, AOL (formerly America On-
line) provided its costumers access to their catalog of sites
and services. With the advent of the World Wide Web and
email, users were no longer bound to the web of a single com-
pany, but could explore the Internet freely and write emails
to any other user using a different ISP. The decentralized
concept of the World Wide Web allowed the development of
open communication platforms.
A more recent example for a decentralized system is
XMPP4(Extensible Messaging and Presence Protocol). The
messaging works similar to email, with users having an iden-
tifier with a server, and servers sending messages to each
other. Additionally, the idea of presence allows users to reg-
ister with their server with different devices. This presence
can be published to the own address book (buddies ).
3http://www.wechat.com/
4http://xmpp.org/
At the same time, the recent past also showed a trend
towards new centralized systems, most prominently Face-
book5. Every few weeks, especially in the EU, Facebook is
in the news because of privacy issues and many users are
concerned [6, 4].
Overall, one major advantage of decentralized systems
is that they offer users the possibility to change service
providers. Interconnection between different providers can
be guaranteed through standardized interfaces.
3.2 Online and Mobile Social Networks
In [14], decentralized OSNs are surveyed. There is a cate-
gorization of architectural approaches that distinguishes be-
tween peer-to-peer (P2P), federated OSNs (F-OSN), and hy-
brid approaches. Going with these definitions, P2P OSNs
leverage peer-to-peer-technology without a central server.
F-OSNs use client/server infrastructure with multiple inter-
connected servers maintained by different providers or peo-
ple. Hybrid approaches combine the two other approaches.
PeerSoN is a P2P OSN that focuses on privacy issues [5].
The project considers offline scenarios by proposing to store
messages for offline users in a distributed hash table (DHT).
Prometheus is another OSN project that utilizes a DHT [12].
A public-key infrastructure is employed, and users can define
access control lists for access to their data. By collecting
data from existing OSNs, emails, etc., weighted and named
edges are constructed in the social graph. Data is replicated
with trusted nodes in the DHT. The nodes should always
be online, and mobile devices are not considered suitable
nodes for the project. Overall, the focus here is both the
preservation of privacy as well as an approximation of the
real social graph.
Diaspora6is an example for a federated OSN. Here, users
can choose between existing servers or set up their own
server. The users benefit from this architectural approach
with the ability to change providers, similar to how it is
possible with email. Privacy concerns are reduced because
no single service provider has all the information about all
users. The user can be in control of his own data, but for the
system as a whole, search and recommendations are more
complicated because of the distributed nature of the data.
According to [14], one advantage of F-OSNs in comparison
with P2P OSNs is the typically better usability.
The main categories that are further analyzed in [14] are
storage, access control, and interaction methods (i.e., com-
munication protocols). According to Paul et al. [14], the
main motivation for research in the field of decentralized
OSNs is privacy concerns with centralized OSNs. Location-
based scenarios and mobile devices are not considered.
Regarding MSNs, in [11], a distinction between three ar-
chitectural approaches is made: centralized,distributed, and
hybrid architectures. A centralized architecture is an archi-
tecture, in which a server delivers data to users on mobile
devices. One example given is Facebook. For such a cen-
tralized architecture, one main advantage is that search and
recommendations are better supported through knowledge
about all users and the whole social graph. For the users
of such a system, there are the disadvantages of the lock-in-
effect and privacy concerns.
With ”distributed” architectures, the authors describe ap-
proaches that do not rely on existing infrastructures, but
5https://www.facebook.com/
6https://diasporafoundation.org/
opportunistic connections between mobile devices via Blue-
tooth or Wi-Fi [11]. One example is MobiClique [15]. Blue-
tooth is used for the transferal of data between users. Mes-
sages are forwarded between peers, instant communication
as well as the arrival of a message cannot be guaranteed.
While especially the uncertain delivery of a message is a
major disadvantage, there are scenarios where such an ap-
proach might be useful, e.g., after a natural catastrophe that
leaves no intact communication infrastructure behind. A hy-
brid architecture combines the two models, centralized and
distributed, by allowing communication between users on
mobile devices utilizing opportunistic connections, while al-
lowing the retrieval of content via central servers.
Kayastha et al. further investigate community detection
for location-based scenarios [11]. The presented approaches
seem to assume knowledge about the whole social graph.
We argue that when utilizing a distributed social graph, the
whole social graph is not known to any one entity and that
new mechanisms based on such a distributed social graph
are needed for community detection and formation.
Bellavista et al. give a classification of MSNs on a different
level in [3]. Here, a distinction is made by the parameters by
which communities can be formed. The classification system
spans pairs of mutually exclusive properties: user-centric
(friends of someone) versus place-centric (e.g., fans of a spe-
cific museum), location-dependent (people at the location)
versus location-independent (people anywhere). Addition-
ally, regarding the storage and analysis of user metadata,
Bellavista et al. distinguish between a spatial scope - by
defining a global (data from anywhere) versus a local (data
stored at the museum only) scope - and a temporal scope
(if user metadata is stored for a longer time or only used
instantaneously). In Section 4, we propose an new approach
to categorize the creation of groups in OSNs.
Regarding the collection of context data on smartphones,
some previous work has been done. With P3 Systems, Jones
and Grandhi describe people-to-people-to-geographical-places
[10]. The authors describe location-based services and what
potential lies in them, especially with respect to online so-
cial networking and recommendations. Ideas of using users’
proximity or location histories for recommender systems are
presented.
In WhozThat?, Bluetooth is used for the broadcasting of
social IDs, e.g., usernames from existing OSNs [1]. The
users’ phones can then retrieve information about other
users in proximity by requesting public profile information
from OSNs. The authors report about an implementation
that leverages this approach to realize a music player that
automatically plays music according to the tastes of the peo-
ple in proximity. Security and privacy issues are mentioned,
but not discussed further in [1].
In SocialFusion, the idea of collection context data is ex-
tended to not only data from existing OSNs, but also from
mobile phones and nearby sensors [2]. In both WhozThat?
and SocialFusion, OSN aspects, like the interaction between
users, is not intended.
4. THREE-TIER VIEW OF THE
SOCIAL GRAPH
In contrast to [3], we propose a new, easier, and in our
eyes more useful approach to categorize group formation in
OSNs. By detailing this approach, we will identify architec-
tural requirements for an OSN architecture. The following
three-tier view of the social graph will be motivated by an
advanced user scenario:
Location-based groups: short-term groups which can
form spontaneously when people are at the same loca-
tion.
Context-based groups: medium-term groups which are
formed based on context, e.g., being in the same class
at university.
Friends: long-term relations which are independent
from context and location.
Using the terminology from [7], the friend-tier connections
represent the self-reported social network, while context- and
location-tier relations represent the inferred or detected so-
cial network.
To further illustrate these three tiers we will look at an-
other user scenario. Imagine a chain of gyms Alice is regis-
tered with and regularly goes to. The machines she works
out with transmit data about her workout to her smart-
phone. Furthermore, her wearable fitness tracker collects
data. While she is working out, she listens to music on her
smartphone. Her top ten artists are always automatically
maintained by her mobile device. From the workout data
on her phone, a fitness index is calculated. As all the data
is on her phone, Alice is in control of her data. Should she
decide to change to another gym chain, she has all her data
with her and can carry settings for workout machines etc.
with her.
Friend-Tier.
To share her fitness index or her top ten favorite music
artists with her friends, Alice first needs to connect to her
friends. What is needed here is a global directory in which
Alice can look up other users by some form of ID or user-
name. Such a DNS-like directory should further contain in-
formation about the presence of the friend, and information
about how to contact this friend, e.g., IP address and proto-
col. Such a global directory can be used independently from
actual OSN applications and would be used for looking up
how to contact people, not storing social graph information.
Connecting to her friends, Alice has formed her part of the
distributed social graph in the friend-tier.
Context-Tier.
On the context-tier, Alice can be added to a group with
the other people that work out at the same gym chain, inde-
pendent of the branch location. As the global directory does
only store information about how to contact other users, an-
other form of directory is needed for creating a group based
on context. The gym chain could offer a Wi-Fi hotspot that
offers its customers to share their ID or username.
Additionally to just sharing IDs to form dynamic context-
tier groups, a gym could have a TV in the lobby where the
users that are currently working out are displayed. Addi-
tionally, the top music artists listened to by the people work-
ing out could be displayed. The users would stay in control
of their data and could decide if and what data they share
when entering the gym. Furthermore, the gym could use the
provided context data to give personalized recommendations
for products that fit a customer’s workout style.
Retrieve Profile
Information
Context-Tier
Location-Tier
Calculate common
contacts and interests
Friend-Tier
Figure 1: Three-tier Social Graph
Location-Tier.
During the summertime, some gym customers might pre-
fer to go jogging outside in a park instead of working out
in a gym. Here, location-tier groups can be formed. Some
trusted online directory, e.g., provided by the gym chain or
the city, is needed for the collection of users that are cur-
rently at the park. When such a directory offers its users the
ability to publish more data than just their IDs, e.g., their
fitness index, the directory can be used for searching, for
example, to find new jogging partners with a similar fitness
index.
5. T3 ARCHITECTURE
The three-tier view developed in the previous section in-
duces requirements for a supporting OSN architecture which
we will summarize in the following. We introduce an archi-
tecture called T3 satisfying these requirements.
To enable users to look up how to contact other users in
decentralized OSN architectures, we propose a global direc-
tory, independent of the concrete OSN application or used
communication protocol. Furthermore, a communication ar-
chitecture that allows for direct communications between
two smartphones over the Internet eliminates the need for a
central server.
Additional decentralized directories can be implemented
that allow users to publish their context data, enabling
search and recommendations, as well as enabling support for
the introduced context- and location-tier of the three-tiered
view of the social graph. Leveraging the smartphone’s capa-
bilities, context data can be collected and processed locally.
Standardized data formats are needed for the exchange be-
tween different users.
In Figure 1, we give an example of a social graph according
to our three-tier approach. The smartphone in the center of
the image is in a location-tier group with other users that
are at the same location, as indicated by the dark gray circle.
The light gray ellipse indicates membership in a context-tier
group based on a common context, for example, the people
that are enrolled in the same class at a university. The friend
tier connections are indicated by the light gray triangle.
1. Send Context and
Location Data
2. Determine
Recommendations
Based on Similar Users
3. Give Recommendations
Figure 2: Recommendation Functionality of a Decentralized
Directory
The black lines indicate direct interactions. The black
line on the right hand side in Figure 1 indicates that the
center smartphone is calculating its common contacts and
interests with another smartphone in the area. The black
line on the left hand side in Figure 1 indicates that from
another smartphone, the center smartphone retrieves profile
information.
Figure 2 shows the general concept of a decentralized di-
rectory. Imagine a travel scenario in which user Alice is
querying a computer she encounters at a central train sta-
tion of the city she is visiting. Alice sends location (history)
and context data to the computer and receives personalized
recommendations based on the data she sent. Additionally
to recommendations, a decentralized directory can be uti-
lized for the search for nearby restaurants or attractions.
In Figure 3, we outline our proposed T3 architecture for
OSNs. In the following, we will detail the role the com-
ponents play by indicating what of the six core features of
OSNs (Section 2.3), they realize. The core components are:
Global Directory.
The global directory is used to look up how to contact
other users for interaction, e.g., calling or messaging. Fur-
thermore, information about where and how to retrieve a
user’s personal profile could be stored in the global direc-
tory as well.
Decentralized Directories.
Decentralized directories support search and recommen-
dations by allowing users registration and by offering the
possibility for users to publish context and location data.
Compared to a single central directory service, one advan-
tage of the approach of having multiple decentralized direc-
tories is trust: users might put higher trust in their gym
chain that they visit regularly and where they see and know
the people working there. Furthermore, users can choose
which data to share with each directory, there is no need for
a central directory that has all the data.
Smartphones.
On the users’ smartphones, the social graph is stored in a
distributed manner. Additionally, each user is in control of
the locally stored context information. The personal profile
could be hosted on the mobile devices, or on an external
server for better availability. Some form of access control
will be necessary to ensure proper privacy settings for users.
For this, the three tiers could be utilized, or some additional
role-based model - with roles like co-workers or classmates
- could be used for access control automatization.
Global Dire ctory
Decentralized
Directory
Decentralized
Directory
Look Up Users
Interaction
Interaction
Publish Context/Location
Register Search
Recommend Create Context-Tier or
Location-Tier Groups
Look Up Users Look Up Users
...
Figure 3: T3 Architecture for OSNs
The presented T3 architecture supports the three-tier con-
cept of the social graph: Users can connect with friends by
looking them up in the global directory; distributed direc-
tories help form context-tier and location-tier groups. The
users’ privacy is protected by utilizing a distributed social
graph and by giving the users the control over their context
data.
6. CONCLUSION AND OUTLOOK
In this paper, we argued that mobile user scenarios are
essential for OSNs. We defined the core features of OSNs,
introduced a three-tier view of the social graph, and pro-
posed the decentralized OSN architecture T3 supporting this
three-tier view, that utilizes a global directory, additional
decentralized directories, and leverages the capabilities of
current smartphones.
Future work includes the question how user IDs are cre-
ated for the global directory, what authentication and autho-
rization mechanisms are feasible, especially regarding reg-
istration, search, and recommendation with decentralized
directories. Additionally, a feasible publish/subscribe mech-
anism should be investigated to enable users to automat-
ically retrieve status updates from their friends. Further-
more, standardized interfaces for the communication be-
tween smartphones, as well as between smartphones and
directories, are to be researched.
7. ACKNOWLEDGMENTS
This work has received funding from the European
Union’s Horizon 2020 research and innovation program un-
der grant agreement No 645342, project reTHINK; and from
project SONIC (grant number 01IS12056), which is funded
as part of the SoftwareCampus initiative by the German
Federal Ministry of Education and Research (BMBF) in co-
operation with EIT ICT Labs Germany GmbH. The authors
would like to thank Beg¨
um ˙
Ilke Zilci for valuable discussions
about user scenarios.
8. REFERENCES
[1] A. Beach, M. Gartrell, S. Akkala, J. Elston, J. Kelley,
K. Nishimoto, B. Ray, S. Razgulin, K. Sundaresan,
B. Surendar, and others. Whozthat? Evolving an
Ecosystem for Context-Aware Mobile Social Networks.
Network, IEEE, 22(4):50–55, 2008.
[2] A. Beach, M. Gartrell, X. Xing, R. Han, Q. Lv,
S. Mishra, and K. Seada. Fusing mobile, sensor, and
social data to fully enable context-aware computing.
In Proceedings of the Eleventh Workshop on Mobile
Computing Systems & Applications, pages 60–65.
ACM, 2010.
[3] P. Bellavista, R. Montanari, and S. K. Das. Mobile
social networking middleware: A survey. Pervasive
and Mobile Computing, 9(4):437–453, Aug. 2013.
[4] A. Bleicher. The anti-Facebook. IEEE Spectrum,
48(6):54–82, 2011.
[5] S. Buchegger, D. Schi¨
oberg, L.-H. Vu, and A. Datta.
PeerSoN: P2P Social Networking: Early Experiences
and Insights. In Proceedings of the Second ACM
EuroSys Workshop on Social Network Systems, SNS
’09, pages 46–52, New York, NY, USA, 2009. ACM.
[6] A. Datta, S. Buchegger, L.-H. Vu, T. Strufe, and
K. Rzadca. Decentralized Online Social Networks. In
Handbook of Social Network Technologies and
Applications, pages 349–378. Springer, 2010.
[7] N. Eagle, A. S. Pentland, and D. Lazer. Inferring
friendship network structure by using mobile phone
data. Proceedings of the National Academy of
Sciences, 106(36):15274–15278, 2009.
[8] M. Falch, A. Henten, R. Tadayoni, and I. Windekilde.
Business models in social networking. In CMI Int.
Conf. on Social Networking and Communities, 2009.
[9] S. G¨
ond¨
or and H. Hebbo. SONIC: Towards seamless
interaction in heterogeneous distributed OSN
ecosystems. In Proc. Wireless and Mobile Computing,
Networking and Communications, pages 407–412.
IEEE, 2014.
[10] Q. Jones and S. Grandhi. P3 systems: putting the
place back into social networks. IEEE Internet
Computing, 9(5):38–46, Sept. 2005.
[11] N. Kayastha, D. Niyato, P. Wang, and E. Hossain.
Applications, Architectures, and Protocol Design
Issues for Mobile Social Networks: A Survey.
Proceedings of the IEEE, 99(12):2130–2158, Dec. 2011.
[12] N. Kourtellis, J. Finnis, P. Anderson, J. Blackburn,
C. Borcea, and A. Iamnitchi. Prometheus:
User-controlled P2P Social Data Management for
Socially-aware Applications. In Middleware 2010,
pages 212–231. Springer, 2010.
[13] M. McLuhan. Understanding Media: The Extensions
of Man. McGraw-Hill, New York, NY, USA, 1964.
[14] T. Paul, A. Famulari, and T. Strufe. A survey on
decentralized Online Social Networks. Computer
Networks, 75, Part A:437–452, Dec. 2014.
[15] A.-K. Pietil¨
ainen, E. Oliver, J. LeBrun, G. Varghese,
and C. Diot. MobiClique: Middleware for Mobile
Social Networking. In Proceedings of the 2nd ACM
Workshop on Online Social Networks, pages 49–54,
New York, NY, USA, 2009. ACM.
... 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. ...
... Other works enable users to broadcast short profiles or manually entered topics of interest over a short distance [2], [9], [10]. There are also automated approaches, which utilize personal context data in the form of user profiles for comparison [1], [5], [11]. Profiling on natural language resources: In the field of Natural Language Processing (NLP), extracting a user's writing style and selection of topics which reflect the author's demographic and psychological properties is referred to as author profiling. ...
... Let T be a texting history of some user A. In T , we drop all messages that have not been written by A resulting in some reference file R. Let V be the vocabulary of R and #V its vocabulary size. We can now derive the user BoW vector (#v) v∈V , where #v is the number of occurrences of the vocabulary word v in R. The user DF vector can then 5 The backend sees the user's k words to compare similarity on. Submitting randomly selected additional words can sufficiently obfuscate this insight. ...
... 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. ...
... Other works enable users to broadcast short profiles or manually entered topics of interest over a short distance [2], [9], [10]. There are also automated approaches, which utilize personal context data in the form of user profiles for comparison [1], [5], [11]. Profiling on natural language resources: In the field of Natural Language Processing (NLP), extracting a user's writing style and selection of topics which reflect the author's demographic and psychological properties is referred to as author profiling. ...
... Let T be a texting history of some user A. In T , we drop all messages that have not been written by A resulting in some reference file R. Let V be the vocabulary of R and #V its vocabulary size. We can now derive the user BoW vector (#v) v∈V , where #v is the number of occurrences 5 The backend sees the user's k words to compare similarity on. Submitting randomly selected additional words can sufficiently obfuscate this insight. ...
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.
... Accessed: 18.7.2017 "vendors, service providers, and other partners who globally support our business" 9 . ...
... Exceptions to this are Friendica 11 and Diaspora, which allow limited exchange between the two networks by implementing the respective other platform's native protocol and allowing basic data portability through manual extraction and re-import of social profile information. 9 Facebook data policy: https://www.facebook.com/about/privacy/. Accessed: 21.5.2017 10 Diaspora: https://diasporafoundation.org/. ...
... The standard was built to allow defining nature and status of a relationship, as well as built-in access control rules. Furthermore, OneSocialWeb proposed standards for Activity Streams and vCard via XMPP, as well as a Personal Eventing Storage for storage and distribution of content 9 . PrPl [144] aims at distributing contents of a user's profile to selected servers and cloud services. ...
Thesis
Full-text available
Online Social Networks (OSN) have become an integral part of our everyday lives. We express ourselves, create and collect content such as images or videos, share content and information with our friends and colleagues, exchange messages, or keep track of what’s happening in the world. Yet, despite social communication being implicitly a distributed, decentralized phenomenon, most OSN services are built in a central, monolithic fashion, concentrating all knowledge and power in one company or organization. This contradicts the idea of the social web, as proprietary and isolated walled gardens keep users from being able to freely choose an OSN platform provider or to effectively control their privacy. In order to mitigate the problem, alternative architectures that distribute control and data to multiple independent services were proposed. Unfortunately, the implicit network effects existing in large OSN services still prevent users from migrating to alternative solutions in significant numbers. Moreover, technical protocols for facilitating holistic and seamless interoperability and furthermore data portability in OSN services do not exist. Ultimately, today’s OSN market is dominated by one single service which has been able to attract a significant amount of users, while a large number of competing services and alternative solutions exist that combine a comparably small number of users. Two main issues have been identified that contribute to the current situation of one OSN service heavily dominating the entire market, being the lack of data portability and interoperability between different OSN services. This work proposes Sonic, a solution that aims to interconnect arbitrary OSN services into one open and heterogeneous federation of OSN services. Sonic introduces an open communication protocol and data formats that are able to facilitate interconnectivity of OSN services. The proposed architecture supports core features implemented in today’s most popular OSN services and facilitates extended functionality through an extensibility framework.
... 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. ...
... In general, our context data model can be helpful for the development of any context-aware service, e.g., in the areas of ubiquitous computing and mobile social networking (Beierle et al. 2015;Beierle 2018). After collecting data, we have to analyze to what extend the quality of the context data varies between the variety of different available Android devices. ...
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.
... In general, our context data model can be helpful for the development of any context-aware service, e.g., in the areas of ubiquitous computing and mobile social networking [2,1]. After collecting data, we have to analyze to what extend the quality of the context data varies between the variety of different available Android devices. ...
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.
... In general, our context data model can be helpful for the development of any context-aware service, e.g., in the areas of ubiquitous computing and mobile social networking [2,1]. After collecting data, we have to analyze to what extend the quality of the context data varies between the variety of different available Android devices. ...
Preprint
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 smartphone 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.
... This implies the use of only small amounts of data and a limited number of necessary data exchanges. A lot of the information that is relevant for the social profile of a user is already available on the smartphone itself [5]. As the needed data is present and connectivity between devices is also given, we focus on the question of how to process and compare the data under the constraints of bandwidth and computing limitations of mobile devices. ...
Preprint
While existing social networking services tend to connect people who know each other, people show a desire to also connect to yet unknown people in physical proximity. Existing research shows that people tend to connect to similar people. Utilizing technology in order to stimulate human interaction between strangers, we consider the scenario of two strangers meeting. On the example of similarity in musical taste, we develop a solution for the problem of similarity estimation in proximity-based mobile social networks. We show that a single exchange of a probabilistic data structure between two devices can closely estimate the similarity of two users - without the need to contact a third-party server.We introduce metrics for fast and space-efficient approximation of the Dice coefficient of two multisets - based on the comparison of two Counting Bloom Filters or two Count-Min Sketches. Our analysis shows that utilizing a single hash function minimizes the error when comparing these probabilistic data structures. The size that should be chosen for the data structure depends on the expected average number of unique input elements. Using real user data, we show that a Counting Bloom Filter with a single hash function and a length of 128 is sufficient to accurately estimate the similarity between two multisets representing the musical tastes of two users. Our approach is generalizable for any other similarity estimation of frequencies represented as multisets.
Presentation
Full-text available
Doctoral Defense Presentation Online Social Networks (OSN) have become an integral part of our everyday lives. We express ourselves, create and collect content such as images or videos, share content and information with our friends and colleagues, exchange messages, or keep track of what’s happening in the world. Yet, despite social communication being implicitly a distributed, decentralized phenomenon, most OSN services are built in a central, monolithic fashion, concentrating all knowledge and power in one company or organization. This contradicts the idea of the social web, as proprietary and isolated walled gardens keep users from being able to freely choose an OSN platform provider or to effectively control their privacy. In order to mitigate the problem, alternative architectures that distribute control and data to multiple independent services were proposed. Unfortunately, the implicit network effects existing in large OSN services still prevent users from migrating to alternative solutions in significant numbers. Moreover, technical protocols for facilitating holistic and seamless interoperability and furthermore data portability in OSN services do not exist. Ultimately, today’s OSN market is dominated by one single service which has been able to attract a significant amount of users, while a large number of competing services and alternative solutions exist that combine a comparably small number of users. Two main issues have been identified that contribute to the current situation of one OSN service heavily dominating the entire market, being the lack of data portability and interoperability between different OSN services. This work proposes Sonic, a solution that aims to interconnect arbitrary OSN services into one open and heterogeneous federation of OSN services. Sonic introduces an open communication protocol and data formats that are able to facilitate interconnectivity of OSN services. The proposed architecture supports core features implemented in today’s most popular OSN services and facilitates extended functionality through an extensibility framework.
Conference Paper
Full-text available
Online Social Networks (OSNs) have become an important part of our everyday online lives. We communicate, share content, and organize meetings and events using OSN platforms. 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. 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 issues addressed within the research project SONIC and present the architectural requirements as well as the architectural concepts of the proposed solution. The project proposes an decentralized and heterogeneous federation of OSN platforms connected via a communication protocol that allows different OSN platforms to seamlessly communicate with each other and gives users the ability to migrate social profiles between platforms on demand.
Article
Full-text available
Facebook was launched in 2004, YouTube in 2005, and Twitter in 2006. They are thus very new service platforms but they already have millions of users and their valuation is counted in billions of US$. This is an extraordinary development, which one will not find in any other line of business, and it illustrate the power of network effects, where the utility of the individual users depends on the presence and usage of the network by other users. The cases described in the present paper are clearly examples of web 2.0 services, where the services provided by the networks are dependent upon content created by the users.
Article
Full-text available
In this paper, we identify mobile social networks as an im-portant new direction of research in mobile computing, and show how an expanded definition of mobile social networks that includes sensor networks can enable exciting new context-aware applications, such as context-aware video screens, mu-sic jukeboxes, and mobile health applications. We offer So-cialFusion as a system capable of systematically integrating such diverse mobile, social, and sensing input streams and ef-fectuating the appropriate context-aware output action. We explain some of the major challenges that SocialFusion must overcome. We describe some preliminary results that we have obtained in implementing the SocialFusion vision.
Article
Full-text available
To address privacy concerns over Online Social Networks (OSNs), we propose a distributed, peer-to-peer approach cou-pled with encryption. Extending the distributed approach by direct data exchange between user devices removes the strict connectivity requirements of web-based OSNs. In order to verify the feasibility of this approach, we designed a two-tiered architecture and protocols that recreate the core fea-tures of OSNs in a decentralized way. This paper focuses on the description of the prototype built for the P2P infrastruc-ture for social networks, as a first step without the encryp-tion part, and shares early experiences from the prototype and insights gained since first outlining the challenges and possibilities of decentralized alternatives to OSNs.
Chapter
Full-text available
Current Online social networks (OSN) are web services run on logically centralized infrastructure. Large OSN sites use content distribution networks and thus distribute some of the load by caching for performance reasons, nevertheless there is a central repository for user and application data. This centralized nature of OSNs has several drawbacks including scalability, privacy, dependence on a provider, need for being online for every transaction, and a lack of locality. There have thus been several efforts toward decentralizing OSNs while retaining the functionalities offered by centralized OSNs. A decentralized online social network (DOSN) is a distributed system for social networking with no or limited dependency on any dedicated central infrastructure. In this chapter we explore the various motivations of a decentralized approach to online social networking, discuss several concrete proposals and types of DOSN as well as challenges and opportunities associated with decentralization.
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.
Article
The convergence of social networking and mobile computing is expected to generate a new class of applications, called Mobile Social Networking (MSN) applications, that will be of significant importance for the coming years. Indeed, MSN enhances the capabilities of more traditional Online Social Networking (OSN) to a great extent by enabling mobile users to benefit from opportunistically created social communities; these communities should be determined not only by common interests or contacts but also by mobility-related context, such as physical location and co-presence. In this paper we precisely define what we intend for MSN applications and overview the primary MSN support solutions available in the current literature and that specifically address the underlying technical challenges, design issues, and emerging middleware guidelines. Our primary goal is to identify engineering design criteria for future MSN middleware solutions, capable to flexibly adapt to different application domains and deployment requirements. To this purpose, we present a novel taxonomy of MSN structures and describe how various existing middleware approaches fit the proposed classification; moreover, the survey takes the opportunity of these descriptions to discuss related middleware design/implementation choices determining specific tradeoffs between expressive power, flexibility, and scalability.
Article
The Diaspora guys, four college kids turned chief engineers of the most-talked-about social networking start-up this year, get a lot of friend requests. They imagine that during the next decade, the Web will evolve from a sea of social networking islands into what many developers are calling the federated social Web one that lets you choose your networking provider, just as you now choose your e-mail provider, and yet still connect with friends who use other services.