Social Pervasive Systems

The Integration of Social Networks and Pervasive Systems

Soumaia Al Ayyat

, Sherif G. Aly

and Khaled A. Harras

Department of Computer Science and Engineering, The American University in Cairo, Cairo, Egypt

School of Computer Science, Carnegie Mellon University, Doha, Qatar

Keywords:

Social Pervasive Systems, Context-awareness, Social Networks, SPS Applications, SPS Challenges.

Abstract:

Sensor technology embedded in smart mobile devices branded such devices as candidates for building inno-

vative context-aware pervasive applications. On a parallel front, the notable evolution in the shape and form

of social networking and their seamless accessibility from mobile devices founded a goldmine of contextual

information. Utilizing an ecosystem that combines both mobile smart devices and a big data like environment

in the form of social networks allows for the creation of an elitist set of services and applications that merge

the two domains. In this paper, and following the footsteps of similar research efforts that attempted to com-

bine both domains, we describe what we label as Social Pervasive Systems that cross-pollinate a mutually

inﬂuential mobile and social world with opportunities for new breeds of applications. We present herein the

evolution of the merger between both worlds for a better understanding. Above and beyond what related work

achieved, we present a set of new services and potential applications that emerge from this new blend, and

also describe some of the expected challenges such systems will face.

1 INTRODUCTION

The recent advancement in mobile technologies and

the enabling network infrastructure have paved the

way for the growth of two prime domains, namely

mobile-based pervasive systems and online social net-

works. Mobile technologies have been posed as an

extremely necessary ingredient in the new rise and

implementation of intelligent context-aware applica-

tions (Bellavista and Helal, 2008) (Lehsten et al.,

2010) (Sambasivan et al., 2009). In parallel, the pop-

ularity of online social networks (OSNs) throughout

the world is leading a tide of great social inﬂuence

(Gay, 2009). With such massive social networks pop-

ulation - there are 1 billion active members on Face-

book by Oct. 2012

-, OSNs possess huge amounts

of social information that present a wealthy source

for research. Such immense information can be a

substantial source of contextual information, thus im-

proving the intelligibility of pervasive systems (Kom-

patsiaris et al., 2010). Similarly, by receiving context

from mobile sensors, social networks can be aware of

user context enabling them to provide more intelligent

social services (Quercia et al., 2010).

While Pervasive Systems and Online Social Net-

Facebook Statistics: http://newsroom.fb.com/

works were progressing simultaneously, we show in

this paper that the lack in the enabling technologies

may have hindered their progress. Both domains

continued to evolve at a slow pace until the end of

the 90s (Satyanarayanan, 2001) and were mainly iso-

lated worlds except for very few commercial applica-

tions such as Lovegety and Humming Bird (Eagle and

Pentland, 2005) that attempted to add a ﬂavor from

both systems for better service. Since 2000, both do-

mains witnessed rapid improvement coinciding with

the advancement in sensor technology, mobile tech-

nology and network infrastructure. Initially, both the

pervasive and social networking domains were aware

of the power of integrating forces, yet few elemen-

tary collaborative research work accommodating both

areas led to the emergence of powerful yet incom-

plete contributions as observed by Baldauf et al. (Bal-

dauf et al., 2007) and Boyd et al. (Boyd and Elli-

son, 2008). Few attempts initiated in 2007 contribut-

ing to the merger of both domains, examining their

co-inﬂuence to ultimately maximize beneﬁts (Beach

et al., 2010) (Quercia et al., 2010).

From our analysis of such merger over time, we

envision a further maturity in the merger domain be-

tween the mobile and social worlds. We manifest a

solid merger in the form of new systems that we call

118

Al Ayyat S., Aly S. and Harras K..

Social Pervasive Systems - The Integration of Social Networks and Pervasive Systems.

DOI: 10.5220/0004306101180124

In Proceedings of the 3rd International Conference on Pervasive Embedded Computing and Communication Systems (PECCS-2013), pages 118-124

ISBN: 978-989-8565-43-3

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

Social Pervasive Systems (SPSs). We deﬁne SPS as

a system that intensively utilizes both primitive and

fused context from both the mobile and social worlds.

Relying on the co-inﬂuence of its ancestors, SPS will

infer signiﬁcant bidirectional user context and prefer-

ences between mobile and social systems, thus lead-

ing to a novel breed of wealthy services and applica-

tions. Our contribution in this paper is broken down

to three main points. First, we deﬁne SPS as a new

research thrust and provide illustrative analysis of the

most prominent work towards its emergence. Sec-

ond, we shed some light on the prominent applica-

tions in which SPS would be of striking importance

followed by a discussion of the key challenges that

need to be addressed by the research community. Fi-

nally, we propose pointers for addressing these chal-

lenges along with some proposed solutions.

2 CONTEXT-AWARE SYSTEMS

AND SOCIAL NETWORKS

Context-Aware Systems is one of the research areas

that signiﬁcantly contributes to the construction of

SPS. In the 90s, commercial, non-standard context-

aware applications emerged, along with the challenge

of building applications that use extremely heteroge-

neous sensors. Thus, the progress and scalability of

context-aware systems was signiﬁcantly hindered. In

the past decade, the coinciding improvement of mo-

bile sensor technology and high bandwidth network

infrastructures revived the research and implementa-

tion of pervasive systems (Baldauf et al., 2007).

In parallel to the progress of pervasive systems,

Online social networks (OSNs) transformed from be-

ing ofﬂine applications or mere contact lists into on-

line social network sites that can be accessed off the

web and through APIs (Boyd and Ellison, 2008). We

spent the effort to visualize a timeline for the evolu-

tion of the merger and present it in Figures 1 and 2.

The ﬁgures emphasize the milestones in the evolution

of context-aware systems and OSNs.

3 SOCIAL PERVASIVE SYSTEMS

Many research attempts target either mobile context-

aware systems or social systems, but few have merged

both to maximize the beneﬁts attained from each sys-

tem as compared to when they are separately utilized.

To support this argument, we traced some publication

trends in this domain. We used Google Scholar to sur-

vey the number of publications since 1975 till 2011

that either did unique research in one of the indicated

areas, or underwent combined research in both areas.

The trends illustrated in Figure 3 indicate a modest

activity in research that combines both mobile con-

text aware systems and social networks. We believe

that the fusion of the context extracted from sensors

readily available in mobile devices, along with the

wealth of information available in OSNs, can pro-

duce a new powerful generation of applications serv-

ing a wide range of domains. In this section, we

brieﬂy trace the progress in technology enabling the

infrastructure motivating research towards this cross-

pollination. We then navigate in time to explore early

attempts of a merger between the two domains fol-

lowed by more recent systematic activities, yet still

unstructured. Finally, we deﬁne SPS and illustrate its

fundamental features.

3.1 The Enabling Technology

Technology advancements in affordable mobile tech-

nology, including higher processing powers, larger

memories, and better displays, paved the way for en-

abling the merger between social and mobile context-

aware systems. Notable advancements also emerged

in mobile networking such as the introduction of

Bluetooth in 1998 until the launch of 4G technology

in 2009. Both lines of advancements enabled the cre-

ation of handheld devices that, to a great extent, con-

tinue to compete in the replacement of larger non-

mobile computational devices. Overall, mobile net-

working has become extremely ubiquitous amongst a

large population of mobile users, thus fostering easier

access to OSNs. Simultaneously, mobile sensor tech-

nology improved the innovation of smart phones ever

since its inception in 2000. Devices continue to be

equipped more and more with smaller affordable sen-

sors, which allows devices to be more aware of am-

bience, and thus encouraging the creation of more in-

telligent applications that promote better usability and

sensitivity to user needs. Such parallel improvements

in sensor technology, devices, and networking infras-

tructure all contributed to a high coupling between so-

cial networks and mobile devices, and motivated the

popularity of mobile social network applications and

the smooth access to the online social network sites.

3.2 The Evolution of SPS

From our observations, there have been several shy

attempts of generating premature SPSs since the late

90s. Beginning from 2007, a new era of closer fu-

sion of both domains started as exempliﬁed in a set of

better social and context-aware contributions.

SocialPervasiveSystems-TheIntegrationofSocialNetworksandPervasiveSystems

119

3.2.1 Early Attempts

The early attempts of implementing primitive social

pervasive applications were mainly commercial prod-

ucts. In 1998, the ﬁrst commercial product for in-

troduction systems (Lovegety in Japan) was launched

(Eagle and Pentland, 2005) where users with match-

ing proﬁles were introduced as they come in proxim-

ity. However, it required special devices and manual

feed to generate a primitive social proﬁle. In 2001

Campus Aware (Gay, 2009) was an early attempt to

fuse mobile sensor technology with social networks

as it imposed the visitor’s impressions of speciﬁc lo-

cations on top of the map of a university campus.

In 2002, mobile devices were introduced in user

collocation pattern detection, and common friends

were sought to introduce unacquainted persons who

share common social patterns. For example, the Ex-

perience Project placed proactive displays in the Ubi-

Comp conference next to tag readers to display pre-

senter talks that match the proﬁles of the users in

proximity with tagged conference badges (Eagle and

Pentland, 2005). User proﬁles, however, were stored

in a local database. In 2004, Social Serendipity used

Bluetooth, staged localized social proﬁle systems,

and proximity information, to achieve some kind of

context-awareness (Eagle and Pentland, 2005). In

2006, Eagle et al. introduced the concept of Real-

ity Mining by using mobile phones as behavioral sen-

sors proving their candidacy for sensing human be-

havior. By analyzing the collected data, they build

a system that infers relationships and senses complex

social systems (Eagle and Sandy, 2006). The majority

of the aforementioned research work does not include

the various forms of context and does not communi-

cate with external social networks.

3.2.2 Recent Merger

Integration of context-aware techniques, sensor tech-

nology, social networks, and mobile technology re-

cently emerged and led to fruitful research in context

Figure 1: The evolution of both social networks and context-aware systems from the 1980s to 2006.

Figure 2: The merger between social networks and context-awareness since 2007.

PECCS2013-InternationalConferenceonPervasiveandEmbeddedComputingandCommunicationSystems

120

fusion. For instance, since 2007, the SOCIALNETS

project studies physical and electronic social net-

works to construct an opportunistic virtual and adap-

tive social network that provides knowledge and con-

tent management to pervasive applications (Mendes,

2008). Other studies propose local social networks

as a means for privacy preservation and trust man-

agement policies in location based systems within op-

portunistic networks (Sapuppo and Sø rensen, 2011).

More recent work incorporates social information re-

trieved from OSNs into context-aware systems as a

new form of contextual information utilized in better

service provisioning (Beach et al., 2010).

The year 2009 has witnessed many contribu-

tions including: Social context-aware browsing to

improve search techniques (Vassena, 2009), apply-

ing the social approach in resolving context-aware

system group conﬂicts (Kwon, 2009) (Beach et al.,

2010), social context-aware services that feed into so-

cial networks (Santos et al., 2009), adaptation and

support for manual control of context-aware systems

(Beach et al., 2010) to gain users’ trust and conﬁ-

dence, social pervasive e-tourism that infers user con-

text and mobility proﬁle to provide recommendation

services (Garcia-Crespo et al., 2009) (Kompatsiaris

et al., 2010), and integrating online user proﬁles with

face-to-face presence (den Broeck et al., 2010). In

2010, a closer fusion among context-awareness and

online social networks, resulted in the following con-

tributions: Context fusion to improve new forms of

sensors such as the calendar (Lovett et al., 2010),

logical sensor generation for individuals/group ac-

tion recommendation (Beach et al., 2010), broker-

based social matching service that supports oppor-

tunistic social networking in DTNs (Mokhtar et al.,

2010), recommending and monitoring social relations

as proposed by FriendSensing and SensingHappiness

(Quercia et al., 2010), and ﬁnally integrating user

experience in virtual worlds with the real world, as

in SecondLife, by feeding current user actions - be-

ing extracted from the mobile sensors - into a virtual

world account (Mahmud et al., 2010).

3.3 Social Pervasive Systems: Deﬁnition

and Features

Social Pervasive Systems are the cross-pollination of

mobile systems and social systems that intensely use

mutual context. They inherit the main features of both

domains, yet are depicted by a new breed of features.

Any SPS inherits from pervasive systems context-

awareness, invisibility, handling data from diverse

sensors, dealing with heterogeneity, context manage-

ment, being proactive, adaptation, security preserva-

Figure 3: Trend of research in social networks, pervasive

computing and joint research since 1975.

tion and scalability (Bellavista and Helal, 2008). On

another front, SPS inherits from social systems so-

cial proﬁle generation, social communication, interest

groups generation, media sharing, posting peer com-

ments, tagging, allowing application sharing, friend-

ship network generation towards more social net-

working, and the provisioning of several levels of pri-

vacy settings (Boyd and Ellison, 2008).

However, SPSs have a unique set of prominent

features - as we envision it - due to the fusion of

signiﬁcant attributes of both ancestors and the co-

inﬂuence of both domains. In such kinds of systems,

social networks will provide both low level and fused

social context to mobile devices using logical social

sensors. On the other side, SPSs can inﬂuence social

networks by exploiting mobile-device-based sensa-

tion to improvethe social networks’ awareness and in-

telligence. To successfully exist, these systems must

manage in real-time the possibly massive amounts of

diverse data to avoid staleness, outdated context infer-

ence, and inappropriate SPS actions and recommen-

dations.

We extrapolate a set of optional features for SPS

performance enhancement such as: Inclusion of cer-

tainty levels with situation identiﬁcation, offering op-

tional user customization of the system at all stages

to improve system ﬂexibility, and possibly utilizing

cloud technology in data storage distribution and pro-

cessing. Finally, SPS may rely on opportunistic net-

works to widen the range of applications and to pro-

vide dynamic adaptation based on communication.

4 SPS APPLICATIONS

With this cross-pollination, a breed of new applica-

tion families emerges to provide a higher level of so-

cially inﬂuencing services. In this section, we de-

scribe ﬁve different families of applications that may

emerge from such kind of merger.

- Monitoring Social Behavior. These systems mon-

itor the cumulative behavioral proﬁles of the social

groups within a certain population in order to cat-

SocialPervasiveSystems-TheIntegrationofSocialNetworksandPervasiveSystems

121

egorize them into sub-populations sharing common

features, economic conditions and health conditions.

These systems consist of two phases; ﬁrst, gathering

the collective behavioral proﬁles from both context-

aware mobile systems and social networks, and then

analyzing and grouping these proﬁles to deduce the

common sub-populations. The behavioral data gath-

ering phase mainly relies on collaborative, large-

scale sensing among people’s mobile phones com-

bined with social information collected from social

systems. There are signiﬁcant behavioral data gath-

ering projects that support the data gathering phase

such as Reality Mining (Eagle and Sandy, 2006) and

”Big Data” (Eagle, 2010) which provide massive so-

cial/behavioral data gathered from mobile phones.

The second phase requires signiﬁcant research con-

tributions in the ﬁeld of high-level context inference

to deduce and analyze the common social proﬁles

among social subgroups.

Such social behavior monitoring systems are sig-

niﬁcant for a wide spectrum of applications such

as population-targeting healthcare applications, eco-

nomic/political applications targeting certain sub-

populations, urban sensing, monitoring trafﬁc con-

gestion and analyzing social interactions. Further-

more, fusing social tagging with context-awareness

and multimedia technology enriches Social Multime-

dia (Tian et al., 2010) which constitutes a wealthy

source for social context-awaresearchengines and so-

cial studies (Kompatsiaris et al., 2010).

- Social Persuasive Applications. Effective and per-

suasive applications such as social pervasive adver-

tising would use both mobile context and social con-

text to achieve many objectives. One of the objectives

could be to propagate customizable and appealing ad-

vertisements. Such applications can persuade clients

to purchase certain kinds of products or services by

having prior knowledge of how such products or ser-

vices relate to the customers’ mobility or social inter-

ests. Further information about social contacts, their

preferences, the degree of their inﬂuence on the sub-

ject in question can even aid the effective persuasion

process. Furthermore, persuading users to move to

certain regions in commercial areas can even have

more of an impact as opposed to simply purchasing

the products or services, such as decongesting physi-

cal areas by routing customers to areas of lower com-

mercial value.

Persuasive systems can also have educational ap-

plications. Amongst many, they can rely on user’s

social proﬁles, current context, and the context of

friends and peers. Fusing all this data can produce

information on the academic progress of users com-

pared to their peers in an act of academic persuasion

to do better. Peers with higher interaction with a user

may even be more inﬂuential than others with less in-

teraction patterns, so ranking social contacts’ interac-

tion could be very useful for applications of this kind.

- Socially Inﬂuenced Context-aware Systems. By

utilizing the history of mobility, social interactions,

social and behavioral proﬁles, such kinds of systems

can deduce their future actions and provide suitable

services. Among the services provided by this fam-

ily of applications is identifying mobile nodes capa-

ble of forwarding information in a way that is sensi-

tive to their behavioral proﬁle and their willingness

to perform such actions. Social-aware ad hoc mes-

sage forwarding approaches is a hot research area that

can beneﬁt from this application family; in which we

propose incorporating interest, awareness of the re-

maining node’s power, willingness to forward mes-

sages, activeness and social ranking among other con-

text and social-aware parameters. Another exam-

ple is social/context-aware search engines that can be

guided by users’ current context, social preferences

and the history of browsing actions.

- Alerting Systems. Alerting systems monitor user

behavioral patterns and gather current user context

from various logical and physical sensors to infer any

changes in their behavior, and to deduce whether such

behavioral changes constitute alerting phenomena. In

case of deduced alerting status, these systems alert

users or their social contacts of the detected critical

status. The alerting systems can infer the proper so-

cial contacts to alert based on the ranking of social

interactions in the social graph.

Alerting systems can be utilized in applica-

tions alerting users of their peers’ current activi-

ties/progress to motivate them towards better perfor-

mance. We extrapolate on the beneﬁt of alerting sys-

tems in improving nations’ economic status by an-

alyzing outcomes of the social behavior-monitoring

systems to pinpoint the sub-populations on the verge

of economic danger. Furthermore, governments may

be alerted by these applications of the sub-populations

in need of direct support to avoid approaching eco-

nomic crises.

- Social Recommender Systems. They are one of

the great potential applications of SPS. They pro-

vide recommendations based on situation/context in-

ferred from the fusion of social information, user and

peers’ context and behavioral proﬁles. Situation/high-

level context inference is an active research area that

heavily relies on context information extracted from

OSNs, mobile devices, logical and physical sensors

combined with machine learning approaches.

We visualize these systems in work environments

where they may recommend to managers the optimal

PECCS2013-InternationalConferenceonPervasiveandEmbeddedComputingandCommunicationSystems

122

group of employees to perform certain tasks based

upon an analysis of social interactions and personali-

ties obtained from their social proﬁles. These applica-

tions can also provide common services within work

premises to all employees based on context. Such ser-

vices can be adapted to mobile device capabilities, or

the current requirements as per the task in hand.

Other types of recommender systems can include

social context-aware learning management systems

that can, for example, customize study material as

per student level of comprehension, recommend help-

ful interest groups, and even persuade students to

progress in their studies through alerts of colleagues’

progress. Such systems can also generate study

groups based on students’ common study habits. On

another front, the application may alert instructors of

those students facing difﬁculty in comprehending cer-

tain lessons, and recommend suitable aiding methods,

and may also refer instructors to suitable references to

improve their teaching skills, all based on information

co-obtained from both pervasive and social worlds.

5 SPS CHALLENGES

To realize our proposed SPS application families, we

list some challenges that need to be addressed by the

research community in order to realize the full poten-

tial of SPS. We believe the following to be the most

prominent set of challenges:

- Intelligence and Invisibility. Given the large spec-

trum of contextual data an SPS can attain from both

the mobile and social worlds, further challenges of

intelligence and invisibility arise. Irrespective of the

amount of sources of contextual information that are

present in such systems, the need for user interven-

tion in processing should be minimized as usual. Is-

sues like the utilization of varying social networks,

authenticating credentials, and diverse mobile plat-

forms should all be invisible to users.

- Extreme Heterogeneity. Above and beyond typ-

ical heterogeneity issues in pervasive systems, SPSs

are challenged by magniﬁed heterogeneity challenges

given they involve both social networks and mobile

systems altogether. Challenges involving access to

various social networks like Facebook, Twitter or

Blackboard include amongst many, varying structures

for the types of context that could be retrieved, and

varying APIs for access. In addition, varying com-

munication infrastructures, differences in interacting

devices, and login credentials of the various interact-

ing social/mobile systems still exist.

- Power Conservation. Although power and energy

are typical challenges in pervasive systems, the huge

information exchange that may be involved in sys-

tems combining the mobile and social worlds imposes

a higher power conservation challenge. Higher com-

munication requirements impose a larger challenge

due to the amount of battery power depletion that will

be incurred upon mobile devices.

- Real-time Constraints. The trade-off between real-

time management of input data and processing time

consumption requires close attention. The rigid time

constraints expected in many SPS applications are

also coupled with the need for huge storage space and

large processing power in order to process large col-

lected contextual data.

- Security and Privacy. Privacy problems still hold

while exchanging and utilizing social and mobile in-

formation, especially when hosts are used in such sys-

tems whose reputation and trust is not well deﬁned.

Unsuitable privacy levels especially when social in-

formation is utilized create the risk of users refraining

from trusting and using such systems.

- Scalability and Conﬂict Resolution. Scalability

is an inevitable challenge that needs further research

efforts. The majority of SPSs offer services in lo-

calized areas. Once the range of services increases,

performance typically deteriorates. Furthermore, pro-

cessing massive noisy real-time data, latency, infer-

ring context from data with degrees of uncertainty are

among the challenges rising from scalability (Cook

and Das, 2012). On another front, the more differ-

ent sources of contextual information are used from

mobile and social networks, conﬂicting contextual in-

formation becomes a notable risk.

6 PROPOSED SOLUTIONS

In this section, we present possible solutions for ad-

dressing some of the aforementioned challenges. To

achieve intelligibility , better situation inference al-

gorithms are needed to properly react towards satis-

fying user needs and expectations, and to gain user

conﬁdence in the intelligibility of applications. Possi-

ble approaches for situation inference would require

combining rules upon multitudes of observed activi-

ties such as recent user social interactions, response

to recent communications and change in routine be-

havioral proﬁles.

To overcome the challenge of heterogeneity, stan-

dardization of context-data exchange among various

systems/sensors becomes more imminent. Such stan-

dardization is achievable through a standard context

markup language or a standard format/protocol for

SocialPervasiveSystems-TheIntegrationofSocialNetworksandPervasiveSystems

123

exchanged context information. Standardized login

credentials are also needed. Initial contributions pro-

pose an ”identity aggregator” to store these creden-

tials and transparently map the services to the proper

credentials (Beach et al., 2010). However, the ”iden-

tity aggregator” idea faces its own set of challenges

such as security threats, ad hoc communication, cen-

tralized versus distributed processing. Furthermore,

research studies are needed to determine the best dis-

tribution of the expected collection of data for satis-

fying optimal processing times and storage and I/O

cost.

For power conservation, some research contribu-

tions propose solutions such as balancing process-

ing and communication among the resource-rich and

resource-weak devices (Schuhmann et al., 2010). The

frequencyof context information update could also be

set based on its frequent change and urgency (Beach

et al., 2010), and trends to use renewable power

sources such as solar power and wind in charging the

SPSs (Cook and Das, 2012) could be used. More in-

novative energy harvesting techniques can be adopted

such as transforming the electromagnetic waves of

the surrounding objects or the negative human en-

ergy (NHE) into an electric form suitable for charging

wearable/mobile devices. On another front, inclusion

of NHE-detection sensors in mobile devices, coupled

with the user status extracted from OSN, can guide

both systems to adapt their themes/applications to the

currently detected negative user mood.

Distributed processing and Cloud computing are

among the promising venues that support SPS scal-

ability and compensate for resource deﬁciency in

resource-weak devices. Cloud computing also shows

potential in resolving the stress of massive real-time

processing. Besides, we believe that predicting con-

text ofﬂine can save time and reduce the stress of the

huge real-time processing.

Finally, for privacy preservation, users may set

context-aware privacy settings that change per loca-

tion, time, or user mood. In addition, secure zones

through which SPS can safely migrate their context

data could be researched.

REFERENCES

Baldauf, M. et al. (2007). A survey on context-aware sys-

tems. Int. J. Ad Hoc and Ubiquitous Computing,

2(4):263–277.

Beach, A. et al. (2010). Fusing mobile, sensor, and social

data to fully enable context-aware computing. In Hot-

Mobile, pages 60–65.

Bellavista, P. and Helal, S. (2008). Location-Based Ser-

vices: Back to the Future. Per. Comp., 7(2):85–89.

Boyd, D. and Ellison, N. (2008). Social Network Sites:

Deﬁnition, History, and Scholarship. J. of Computer-

Mediated Communication, 13(1):210–230.

Cook, D. and Das, S. (2012). Pervasive computing at scale:

Transforming the state of the art. Pervasive and Mo-

bile Computing, 8(1):22–35.

den Broeck, W. et al. (2010). The Live Social Semantics ap-

plication: a platform for integrating face-to-face pres-

ence with on-line social networking. In PERCOM

Workshops, pages 226–231.

Eagle, N. (2010). Big data, global development, and com-

plex social systems. In ACM SIGSOFT, pages 3–4.

Eagle, N. and Pentland, A. (2005). Social Serendipity: Mo-

bilizing Social Software. Per. Comp., 4:28–34.

Eagle, N. and Sandy, A. A. (2006). Reality mining : sens-

ing complex social systems. Personal and Ubiquitous

Computing, 10(4):255–268.

Garcia-Crespo, A. et al. (2009). SPETA: Social perva-

sive e-Tourism advisor. Telematics and Informatics,

26(3):306–315.

Gay, G. (2009). Context-Aware Mobile Computing: Affor-

dances of Space, Social Awareness, and Social Inﬂu-

ence. Morgan & Claypool.

Kompatsiaris, Y. et al. (2010). Information Extraction from

Social Sites. In SSMS.

Kwon, O. (2009). A social network approach to resolving

group-level conﬂict in context-aware services. Expert

Systems with Applications, 36(5):8967–8974.

Lehsten, P. et al. (2010). A Service-oriented Approach to-

wards Context-aware Mobile Learning Management

Systems. In PERCOM, pages 268–273.

Lovett, T. et al. (2010). The Calendar as a Sensor: Anal-

ysis and Improvement Using Data Fusion with Social

Networks and Location. In UbiComp, pages 3–12.

Mahmud, J. et al. (2010). AVARA: A system to improve

user experience in web and virtual world. In IUI Pro-

ceedings, pages 349–352.

Mendes, J. (2008). SOCIALNETS: Social networking for

pervasive adaptation. Technical Report 217141, EC.

Mokhtar, S. B. et al. (2010). A self-organising directory and

matching service for opportunistic social networking.

In SNS Workshop, pages 1–6.

Quercia, D. et al. (2010). Using Mobile Phones to Nurture

Social Networks. Per. Comp., 9(3):12–20.

Sambasivan, N. et al. (2009). UbiComp4D: Infrastruc-

ture and Interaction for International Development-the

Case of Urban Indian Slums. In ACM UbiComp.

Santos, A. et al. (2009). Context Inference for Mobile Ap-

plications in the UPCASE Project. In MOBILe Wire-

less MiddleWARE - MOBILWARE, pages 352–365.

Sapuppo, A. and Sø rensen, L. (2011). Local Social Net-

works. In ICTTA, volume 5, pages 15–22.

Satyanarayanan, M. (2001). Pervasive computing: vision

and challenges. Personal Communications, IEEE,

8(4):10–17.

Schuhmann, S. et al. (2010). Efﬁcient Resource-Aware Hy-

brid Conﬁguration of Distributed Pervasive Applica-

tions. In PERCOM, pages 373–390.

Tian, Y. et al. (2010). Social Multimedia Computing. IEEE

Computer Society, 43(8):27–36.

Vassena, L. (2009). Context-aware retrieval going social. In

FDIA Proceedings, pages 62–68.

PECCS2013-InternationalConferenceonPervasiveandEmbeddedComputingandCommunicationSystems

124