ISSN (Online): 2349-7084

GLOBAL IMPACT FACTOR 0.238

ISRA JIF 0.351
INTERNATIONAL JOURNAL OF COMPUTER ENGINEERING IN RESEARCH TRENDS
VOLUME 1, ISSUE 6, DECEMBER 2014, PP 384-390

Mobile Social Networks for Flattering

Unsigned Profile Matching
1
1

ANUSHA.V, 2 K.SUMALATHA

M.Tech Research Scholar, Priyadarshini Institute of Technology and Science for Women
2
Assistant Professor, Priyadarshini Institute of Technology and Science for Women

Abstract: Social networking makes digital communication technologies sharpening tools for extending the social circle of
people. Privacy preservation is a significant research issue in social networking. Here user Profile matching with privacy
preservation in mobile social networks (MSNs) is studied and a family of Profile matching protocols is introduced. An explicit
Comparison-based Profile matching protocol (eCPM) which runs between two parties, an initiator and a responder is
proposed which enables the initiator to obtain the comparison-based matching result about a specified attribute in their
Profiles, while preventing their attribute values from disclosure. An implicit Comparison-based Profile matching protocol
(iCPM) is then proposed which allows the initiator to directly obtain some messages instead of the comparison result from
the responder. The messages unrelated to user Profile can be divided into multiple categories by the responder. The initiator
implicitly chooses the interested category which is unknown to the responder. Two messages in each category are prepared
by the responder, and only one message can be obtained by the initiator according to the comparison result on a single
attribute. iCPM is further generalized into an implicit Predicate-based Profile matching protocol (iPPM) which allows complex
comparison criteria spanning multiple attributes.

Keywords: Mobile social network, Profile matching, privacy preservation, Homomorphic encryption, oblivious transfer

I. INTRODUCTION

and data devices over networks including cellular

Social networking is where individuals with similar

technology and elements of private and public IP

interests connect with each other through their

infrastructure (such as the Internet). Mobile Social

mobile/tablet. They form virtual communities. For

Networking(MSN) refers to all of the enabling elements

example Face book, Twitter, LinkedIn etc. What makes

necessary for the contribution (posting and uploading)

social network sites unique is not that they allow

and consumption (viewing/experiencing) of social

individuals to meet strangers, but rather that they

media across a mobile network. Key to the definition is

enable users to articulate and make visible their social

the users implicit or explicit choice of network

networks. On many of the large SNSs, Participants are

technologies.

not necessarily Networking or looking to meet new

people; instead they are primarily communicating with

In the MSNs, we consider a generalized function to

the people who are already part of their extended social

support information exchange by using Profile matching

network. To emphasize this articulated social network

as a metric. Following the previous example, we

as a critical organizing feature of these sites, we label

consider two CIA agents with two different priority

them social network sites. Some web based SNSs

levels in the CIA system, A with a low priority lA and B

support limited mobile interactions (e.g... Face book,

with a high priority lB. They know each other as a CIA

MySpace and Cyworld).

agent. However, they do not want to reveal their

Mobile Social Networks is a means of transmitting

priority levels to each other. B wants to share some

information (communicating) using a Mixture of voice

messages to A. The messages are not related to user

IJCERT2014
www.ijcert.org

384

ISSN (Online): 2349-7084

GLOBAL IMPACT FACTOR 0.238
ISRA JIF 0.351
INTERNATIONAL JOURNAL OF COMPUTER ENGINEERING IN RESEARCH TRENDS
VOLUME 1, ISSUE 6, DECEMBER 2014, PP 384-390

Profile, and they are divided into multiple categories,

contrast, our protocols support fine-grained personal

e.g., the messages related to different regions (New York

profiles and thus much finer user differentiation, which

or Beijing) in different years (2011 or 2012). B shares one

is important for fostering the much wider use of PMSN.

message of a specified category at a time.

To our best knowledge, Dong et al. presented the only

piece of work in [5] that does not match two users in

2. RELATED WORK

PMSN using the intersection (cardinality) of their

In general, the Profile matching can be categorized

attribute sets. Instead, they proposed using the social

based on the formats of proles and the types of

proximity between two users as the matching metric,

matching operations. A well-known profile matching is

which measures the distance between their social

the FNP scheme [19], where a client and a server

coordinates with each being a vector precompiled by a

compute their intersection set such that the client gets

trusted central server to represent the location of a user

the result while the server learns nothing. Later, Kissner

in an online social network. By comparison, our work

et al. [31] implemented profile matching with more

does not rely on the affiliation of PMSN users with a

operations including set intersection, union, cardinality

single online social network and addresses a more

and over-threshold operations. On the other hand, Ye et

general private matching problem for PMSN by

al. [32] further extended the FNP scheme to a distributed

supports fine-grained personal profiles and a wide

private matching scheme and Dachman-Soled et al. [33]

spectrum of matching metrics.

aimed at reducing the protocol complexity. All the

In mobile social networking applications,

above solutions to the

set intersection rely on

matching acts as a critical initial step to help users,

homomorphic encryption operation. In the meantime,

especially strangers, initialize conversation with each

other works [34], [35] employed an oblivious pseudo

other in a distributed manner. Yang et al. [30]

random function to build their profile matching

introduced a distributed mobile communication system,

protocols, where communication and computational

called

efficiency is improved. Li et al. [21] implemented profile

networking in physical proximity. E- Small Talker

matching according to three increasing privacy levels: i)

automatically discovers and suggests common topics

revealing the common attribute set of the two users; ii)

between users for easy conversation. Lu et al. [20]

revealing the size of the common attribute set; and iii)

studied e-healthcare cases by proposing a symptom

revealing the size rank of the common attribute sets

matching scheme for mobile health social networks.

between a user and its neighbors. They considered an

They considered that such matching scheme is valuable

honest-but-curious (HBC) adversary model, which

to the patients who have the same symptom to exchange

assumes that users try to learn more information than

their experiences, mutual support, and inspiration with

allowed by inferring from the profile matching results,

each other.

E-Small

Talker,

which

facilitates

profile

social

but honestly following the protocol. They applied secure

multi- party computation, the Shamir secret sharing

The proposed profile matching protocols are novel since

scheme, and the homomorphic encryption scheme to

the comparison of attribute values is considered as the

achieve the condentiality of user profiles.

matching operation. The intuitive idea is inspired by the

famous Yaos millionaires problem *37+ and its solution

The private matching schemes proposed in [2][4] aim

[40]. Similar to other works [21][23], we propose three

at coarse-grained personal profiles and match two users

different protocols with different anonymity levels. For

based on a privacy-preserving computation of the

the eCPM with conditional anonymity, we provide

intersection (cardinality) of their attribute sets. In

detailed anonymity analysis and show the relation

IJCERT2014
www.ijcert.org

385

ISSN (Online): 2349-7084

GLOBAL IMPACT FACTOR 0.238
ISRA JIF 0.351
INTERNATIONAL JOURNAL OF COMPUTER ENGINEERING IN RESEARCH TRENDS
VOLUME 1, ISSUE 6, DECEMBER 2014, PP 384-390

between pseudonym change and anonymity variation.

one is assigned to the professional email of the same

For the iCPM and the iPPM with full anonymity, we

user.

show that the use of these protocols does not affect user
anonymity level and users are able to completely
preserve their privacy.

3.2 User Profile Matching in Social

Networks
Inter social network operations and functionalities are

3. PROFILE MATCHING:

required in several scenarios (data integration, data

Profile matching means two users comparing their

enrichment, information retrieval etc...). To achieve this,

personal profiles and is often the first step towards

matching user profiles is required. Profile Matching can

effective PMSN. It, however, conflicts with users

be done using following components.

growing

privacy

concerns

about

disclosing

their

personal profiles to complete strangers before deciding

Components

to interact with them.

FOAF (Friend of a Friend): is admitted to be one of the

real success stories of the semantics web and is
becoming a de facto standard with more and more social

3.1 Matching user profiles on social

networks suffers currently of three main
problems:

networks and tools that allow create/generate FOAF

profiles.
Similarity

Social network offer to users interesting means and

ways to connect communicate and share information
with other members within their platforms. However
those sites have currently different structures/schemas
and they represent users profiles differently. Thus, they
the

exchange

of

information

Assignment:

Comparing

two

profiles comes down to compare (a set of) their

Social Network Representations:

prohibit

Function

and

communication with functioning as Data Isolated

Islands.

attributes. In order to obtain appropriate results,

adapted similarity functions must be associated to each
attribute (e.g.., comparing emails must be computed in a
different way than comparing interests). Various
techniques can be used to measure the similarities score
between two textual/string values.Attribute Weight
Assignment: This component mainly aims to assign a
weight to each attribute in the FOAF vocabulary. This
allows representing the attribute importance within a
defined context. In this framework, the weight can be

User Profile Domains:

Even when sites share the same representation, user
profile attribute domains are not always common. For
instance, the domain values of interests attribute in face
book do not necessarily meet the domain values of the
same attribute in LinkedIn.

assigned manually or computed automatically.

Profile Matcher: This component aims to provide a
decision whether two input profiles refer to the same
physical person or not. Here, two profiles are considered
as representing the same user if their profile similarity
score is higher than a threshold called the profile
matching threshold.

Site/User Objective:
Depending on the site and on the user objectives, the
same attribute can be filled up with two different values.
For instance, the email attribute in face book is
commonly filled with a personal email while LinkedIn

Acquiring the Data: To match user profiles from

different OSN sites, a large and suitable dataset from
social networks is required [12]. The data on the profile
pages is retrieved using a crawler. The data on social
networking sites can be very diverse, unstructured, and

IJCERT2014
www.ijcert.org

386

ISSN (Online): 2349-7084

GLOBAL IMPACT FACTOR 0.238
ISRA JIF 0.351
INTERNATIONAL JOURNAL OF COMPUTER ENGINEERING IN RESEARCH TRENDS
VOLUME 1, ISSUE 6, DECEMBER 2014, PP 384-390

even unsuitable, thus it will need pre-processing. Based

TCA

outputs

certpidi

as

tuple

(pki,

on the knowledge of the structure of the user profile

SignskTCA(pidi,pki)). The homomorphic secret key ski

page and the users friends page in a particular

is delivered to ui together with pski; pki is tied to pidi

network, content can be extracted. Then, irrelevant data

and varies as the change of pseudonyms.

can be filtered out.

Vector Space Model: In the vector space model, both
documents and profiles are represented as vectors with
components for different terms (term vectors) [12].

5. IMPLICIT COMPARISON BASED

APPROACH
Here the implicitbased profile matching (iCPM) is

These components are weights that reflect the frequency

proposed

of each term in the document and interest in a given

cryptographic technique. It is considered that users have

term in the profile, respectively

distinct values for any given attribute. The iCPM

by

adopting

the

oblivious

transfer

consists of three main steps. Then encrypt the vector by

4. EXPLICIT COMPARISON BASED

APPROACH

using the homomorphic encryption and sends the

encrypted vector but still can process on the cipher text.

eCPM protocol allows two users to compare their

attribute values on a specified attribute without

In the second step, computes the cipher texts with input

of self-defined messages for 1 message length.

disclosing the values to each other. But, the protocol

reveals the comparison result to the initiator, and

A. Protocol Steps

therefore offers conditional anonymity. The protocol has

a fundamental bootstrapping phase, where the TCA
generates all system parameters, user pseudonyms, and
keying materials.

A. Bootstrapping
The protocol has a fundamental bootstrapping phase,
where the TCA generates all system parameters, user
pseudonyms, and keying materials. Specifically, the
TCA runs G to generate p, q,R,Rq,Rp, for initiating
Fig 1. ICPM flow

the homomorphic encryption The TCA generates a pair

of public and private keys (pkTCA, skTCA) for itself.
The public key pkTCA is open to all users; the private

6. PERFORMANCE ANALYSIS

key skTCA is a secret which will be used to issue

The performance of three proposed protocols eCPM,

certificates for user pseudonyms and keying materials,

iCPM, and iPPM is studied in terms of communication

as shown below. The TCA generates disjoint sets of

overhead and anonymity strength. When analyzing

pseudonyms (pidi) and disjoint sets of homomorphic

anonymity, it is considered that users have distinct

public keys (pki) for users (ui). For every pidi and pki of

values for any given attribute. Non-distinct attribute

ui, the TCA generates the corresponding secret keys

values and comparison operations and will be

pski and ski. In correspondence to each pseudonym

considered in the future work.

pidi, it assigns a certificate certpidi to ui, which can be

used to confirm the validity of pidi. Generally, the TCA

A. Communication Overhead

uses skTCA to generate a signature on pidi and pki. The

IJCERT2014
www.ijcert.org

387

ISSN (Online): 2349-7084

GLOBAL IMPACT FACTOR 0.238
ISRA JIF 0.351
INTERNATIONAL JOURNAL OF COMPUTER ENGINEERING IN RESEARCH TRENDS
VOLUME 1, ISSUE 6, DECEMBER 2014, PP 384-390

Let |R| be the size of one ring element in Rq. In the

change for consecutive eCPM runs is analyzed. Further

eCPM, the initiator and the responder both need to send

an enhanced version of the eCPM, i.e., eCPM+ is

cipher texts in size of 2|R| and the communication

introduced, by exploiting the prediction-based strategy

overhead is thus subject only to the system parameter

and adopting the pre-adaptive pseudonym change. The

|R|. In order to achieve full anonymity, the iCPM

effectiveness of the eCPM+ is validated through

constructs cipher text in a sequence of operations. It is

extensive

known that |Enc (b)| = 2|R|. Thus, the communication

protocols with full anonymity, i.e., implicit Comparison-

overhead of the initiator is 2( + )|R| with = log l . It

based Profile Matching (iCPM) and implicit Predicate-

can be seen that the initiators communication overhead

based Profile Matching (iPPM) has been devised. The

increases with system parameters (, ). An addition

iCPM handles profile matching based on a single

operation of homomorphic encryption does not increase

comparison

the cipher text size, while a multiplication with inputs of

implemented with a logical expression made of multiple

two ciphertexts of lengths a|R| and b|R| outputs a

comparisons spanning multiple attributes. The iCPM

(a+b1)|R|-length cipher text. Thus, in the iCPM, the

and the iPPM both enable users to anonymously request

communication overhead of the responder increases to

for messages and respond to the requests according to

6|R|. It is concluded that the communication overhead

the profile matching result, without disclosing any

of the eCPM and the iCPM are constantly dependent on

profile information. In current version of the iCPM and

system parameters (, ). The iPPM extends the iCPM

the iPPM, > and < operations for profile matching

by building complex predicates. From the protocol

is implemented.

simulations

of

an

using

attribute

real-trace

while

the

data.

Two

iPPM

is

description, we observe that if a predicate includes n 1

comparisons, the communication overhead of the iPPM

REFERENCES

would be approximately n times of that in the iCPM.

[1]Comscore, http://www.comscoredatamine.com/.

B. Anonymity

[2] A. G. Miklas, K. K. Gollu, K. K. W. Chan, S. Saroiu, P.

Suppose that user ui is currently using pseudonym pidi

K. Gummadi, and E. de Lara, Exploiting social

to execute profile matching with others. An adversary

interactions in mobile systems, in Ubicomp, 2007, pp.

aiming to break the k-anonymity of ui is considered [11].

409428.

K-anonymity is a classic concept for evaluating

[3] S. Ioannidis, A. Chaintreau, and L. Massoulie,

anonymity. It implies that a series of comparison results

Optimal and scalable distribution of content updates

provide k-anonymity protection to a user if the users

over a mobile social network, in Proc. IEEE INFOCOM,

behavior cannot be distinguished from at least k 1

2009, pp. 14221430.

other users.

*4+ R. Lu, X. Lin, and X. Shen, Spring: A social-based

privacy-preserving packet forwarding protocol for

7. CONCLUSION

vehicular delay tolerant networks, in Proc. IEEE

A unique comparison-based profile matching problem

in

Mobile

been

[5] W. He, Y. Huang, K. Nahrstedt, and B. Wu,

investigated, and novel protocols are proposed to solve

Message propagation in ad- hoc-based proximity

it. The explicit Comparison based Profile Matching

mobile social networks, in PERCOM workshops, 2010,

(eCPM) protocol provides conditional anonymity. It

pp. 141146.

reveals

initiator.

[6] D. Niyato, P. Wang, W. Saad, and A. Hjrungnes,

Considering the k-anonymity as a user requirement; the

Controlled coalitional games for cooperative mobile

the

Social

Networks

comparison

result

(MSNs)

to

the

has

INFOCOM, 2010, pp. 632640.

anonymity risk level in relation to the pseudonym

IJCERT2014
www.ijcert.org

388

ISSN (Online): 2349-7084

GLOBAL IMPACT FACTOR 0.238
ISRA JIF 0.351
INTERNATIONAL JOURNAL OF COMPUTER ENGINEERING IN RESEARCH TRENDS
VOLUME 1, ISSUE 6, DECEMBER 2014, PP 384-390

social networks, IEEE Transactions on Vehicular

Balfanz, G. Durfee, N. Shankar, D. K. Smetters, J.

Technology, vol. 60, no. 4, pp. 18121824, 2011.

Staddon, and H.- C. Wong, Secret handshakes from

[7] M. Motani, V. Srinivasan, and P. Nuggehalli,

pairing-based key agreements, in IEEE Symposium on

Peoplenet:

Security and Privacy, 2003, pp. 180196. [19] M. J.

engineering

wireless

virtual

social

network, in MobiCom, 2005, pp. 243257.

Freedman, K. Nissim, and B. Pinkas, Efcient private

[8] M. Brereton, P. Roe, M. Foth, J. M. Bunker, and L.

matching and set intersection, in EUROCRYPT, 2004,

Buys, Designing participation in agile ridesharing with

pp. 119.

mobile social software, in OZCHI, 2009, pp. 257260.

*20+ R. Lu, X. Lin, X. Liang, and X. Shen, A secure

*9+ E.Bulut and B.Szymanski, Exploiting friendship

handshake

relations for efficient routing in delay tolerant mobile

healthcare social network, ACM Mobile Networks and

social networks, IEEE Transactions on Parallel and

Applications (MONET), vol. 16, no. 6, pp. 683694, 2011.

Distributed Systems, vol. 23, no. 12, pp. 22542265, 2012.

*21+ M. Li, N. Cao, S. Yu, and W. Lou, Findu: Privacy-

[10] Z. Yang, B. Zhang, J. Dai, A. C. Champion, D. Xuan,

preserving personal profile matching in mobile social

and D. Li, E-smalltalker: A distributed mobile system

networks, in Proc. IEEE INFOCOM, 2011, pp. 2435

for social networking in physical proximity, in ICDCS,

2443.

2010, pp. 468477.

*22+ R. Zhang, Y. Zhang, J. Sun, and G. Yan, Fine-

*11+ Q. Li, S. Zhu, and G. Cao, Routing in socially

grained private matching for proximity-based mobile

selsh

social networking, in Proc. IEEE INFO- COM, 2012, pp.

delay

tolerant

networks,

in

Proc.

IEEE

scheme

with

symptoms-matching

for

INFOCOM, 2010, pp. 857865.

19691977.

[12] X. Liang, X. Li, T. H. Luan, R. Lu, X. Lin, and X.

*23+ W. Dong, V. Dave, L. Qiu, and Y. Zhang, Secure

Shen, Morality- driven data forwarding with privacy

friend discovery in mobile social networks, in Proc.

preservation

IEEE INFOCOM, 2011, pp. 1647 1655.

in

mobile

social

networks,

IEEE

Transactions on Vehicular Technology, vol. 7, no. 61, pp.

[24] J. Freudiger, M. H. Manshaei, J.-P. Hubaux, and D.

32093222, 2012.

C. Parkes, On non- cooperative location privacy: a

*13+ R. Gross, A. Acquisti, and H. J. H. III, Information

game-theoretic analysis, in ACM CCS, 2009, pp. 324

revelation and privacy in online social networks, in

337.

WPES, 2005, pp. 7180.

[25] R. Lu, X. Lin, H. Luan, X. Liang, and X. Shen,

[14+ F. Stutzman, An evaluation of identity-sharing

Pseudonym changing at social spots: An effective

behavior in social network communities. iDMAa

strategy

Journal, vol. 3, no. 1, pp. 1018, 2006.

Transactions on Vehicular Technology, vol. 61, no. 1, pp.

[15] K. P. N. Puttaswamy, A. Sala, and B. Y. Zhao,

86 96, 2011.

Starclique:

*26+ J. Katz, A. Sahai, and B. Waters, Predicate

guaranteeing

user

privacy

in

social

for

location

privacy

2009, pp. 157168.

equations, and inner products, in EUROCRYPT, 2008,

*16+ E. Zheleva and L. Getoor, To join or not to join: the

pp. 146162.

illusion of privacy in social networks with mixed public

*27+ N. Eagle and A. Pentland, Social serendipity:

and private user proles, in WWW, 2009, pp. 531540.

mobilizing

*17+ G. Chen and F. Rahman, Analyzing privacy

Computing, vol. 4, no. 2, pp. 2834, 2005.

designs of mobile social networking applications,

*28+ J. Teng, B. Zhang, X. Li, X. Bai, and D. Xuan, E-

IEEE/IFIP International Conference on Em- bedded and

shadow: Lubricating social interaction using mobile

Ubiquitous Computing, vol. 2, pp. 8388, 2008. [18] D.

phones, in ICDCS, 2011, pp. 909918.

IJCERT2014

soft-

junctions,

IEEE

encryption

social

dis-

vanets,

networks against intersection attacks, in CoNEXT,

www.ijcert.org

supporting

in

ware,

IEEE

polynomial

Pervasive

389

ISSN (Online): 2349-7084

GLOBAL IMPACT FACTOR 0.238
ISRA JIF 0.351
INTERNATIONAL JOURNAL OF COMPUTER ENGINEERING IN RESEARCH TRENDS
VOLUME 1, ISSUE 6, DECEMBER 2014, PP 384-390

*29+ B. Han and A. Srinivasan, Your friends have more

friends than you do: identifying inuential mobile users
through random walks, in MobiHoc, 2012, pp. 514.
[30] Z. Yang, B. Zhang, J. Dai, A. C. Champion, D. Xuan,
and D. Li, E-smalltalker: A distributed mobile system
for social networking in physical proximity, in ICDCS,
2010, pp. 468477.
*31+ L. Kissner and D. X. Song, Privacy-preserving set
operations, in CRYPTO, 2005, pp. 241257. [32] Q. Ye,
H. Wang, and J. Pieprzyk, Distributed private
matching and set operations, in ISPEC, 2008, pp. 347
360.
[33] D. Dachman-Soled, T. Malkin, M. Raykova, and M.
Yung, Efcient robust private set intersection, in
ACNS, 2009, pp. 125142.
*34+ S.

Jarecki

and X.

Liu,

Efcient

oblivious

pseudorandom function with applications to adaptive ot

and secure computation of set intersection, in TCC,
2009, pp. 577594.
[35] C. Hazay and Y. Lindell, Efcient protocols for set
intersection and pattern matching with security against
malicious

and

covert

adversaries,

Journal

of

Cryptology, vol. 23, no. 3, pp. 422456, 2010.

[36] B.

Goethals,

S.

Laur,

H.

Lipmaa,

and T.

Mielikainen, On private scalar product computation

for privacy-preserving data mining, in ICISC, 2004, pp.
104120.

IJCERT2014
www.ijcert.org

390