Is that really you : An Approach to Assure identity without revealing realname online'

1
Is that really you? An Approach to Assure
identity without revealing real-name online.
DIM2009 Workshop Nov.13 2009

• Akiko ORITA
• Assistant Professor, Chuo University
• ako_at_wide.ad.jp
• Hisakazu HADA
• Assistant Professor, Keio University
• hada_at_sfc.wide.ad.jp

2
Overview

• Illustrate the problem of impersonation in the
  usage of Social Media
• Classify the structure of online identity and its
  representation
• Propose an empirical approach focused on
  bi-directional linkability

3
Background Usage of Pseudonymity

• Over 70 of Internet Users are joining at least
  one online community, but 9.7 of users display
  their real names.
• (Internet White Paper,2007)
• 50 of users of Social Media prefer using
  pseudonyms while 25 prefer slightly
  identified or anonymous.

25
50
25
Data( Orita,2009)
4
When we search friends..

• By Email Address
• Regardless of various screen-name

• By Name or Screen-name
• Regardless of knowing contact information

How do we distinguish whether the account is
your friend or not ?
5
Aggregation and Online Identity

• Lifestream
• By IDs (e.g. email)

• Collection via Website
• By Name/Screen-name
• By relationship

6
Impersonation Pretend to be someone

• Identity theft
• Violation of ID/Password
• Unauthorized access

• Impersonation by name
• By Validly acquired account

Identity will be poisoned by impersonation
7
Problem

• Fake Accounts in Social Media
• Famous People
• Fake Kanye West fake tweets on twitter
• Fake Dalai Lama someone created the account
• General Users
• Fake profiles on Facebook
• It occurs by not only identity theft but also by
  validly acquired account
• How can we figure out the originality?

8
Elements of online identity

• (Un)-Linkability
• the attacker cannot sufficiently distinguish
  whether it exists or not
• The distinction whether the activities of users
  are related or not
• Identity Representation Layer
• Layered Viewpoint of ID and its representation

Anonymity, Unlinkability, Undetectability,
Unobservability,Pseudonymity, and Identity
Management A Consolidated Proposal for
Terminology (Pfitzmann Hansen,2008)
9
Linkability and Unlinkability
Lower Anonymity
Higher Anonymity
Linkable
Unlinkable
message
message
message
message
message
Link
message
message
message
message
Pseudonyms will be given to each user(message
source)
(Pfitzmann et al,1999-2008), Clauß et al (2005)
10
Identity Representation Layers
RealName
Pseudonymity
Anonymity
Users Transaction Layer
Registration
No registration
Service Registered Layer
IP Address
Email Address
Identification Layer
Credential Info
SSN
(Kobsa, et al,2003) (Orita,2009)
11
Pseudonymity and Fakename
Pseudonymity
Anonymity
Users Transaction Layer
FakeName
Registration
No registration
Service Registered Layer
IP Address
Email Address
Identification Layer
Credential Info
SSN
12
Identify the originality

• Traceability to a specific user by basic
  information
• Needless disclosure of personal information
  causes to identity theft
• Linkability of related information
• Detection by similar name

13
Experiment

• Bi-directional links on Twitter Public Timeline
• Period Jun17,2009 1315-1802
• 642 posts from public timeline on Twitter
• Samples were Collected Via Twitter API
• Redundancy of users, independent
• The existence of URL to and from Twitter Profile
• 2. Barack Obama Accounts

14
1. Public timeline (N642)
16 of all the collected TLs are filtered to be
original ones
15
Breakdown of links
16
2. Barack Obama
17
Results

• Bi-directional links
• Considered to be original accounts
• Possibly created by the same user or know each
  other
• Uni-directional links No return links
• Suspicious
• Uni-directional links Different URL backlinks
• Suspicious
• No URLs
• Suspicious ones are filtered out by existence of
  links, without revealing identifier (e.g. email
  address)

18
Discussion

• Bi-directional links should extend to multiple
  platform
• More social cues are collected, more possibility
  of identification appears
• Linkability among observed data
• Avoid modification of data

Sensor
GPS
Exif
19
Conclusion Future Work

• Assure originality of the user
• Focused on bi-directional linkability
• Would be extended to multiple links and observed
  data
• Guideline for authorize Official account
• Candidate for Election
• Public Announcement

Acknowledgement Professional Blogger Dr.
Akimichi OGAWA Internet Business Innovation
Consortium, Keio Univ. Grant-in-Aid for Young
Scientists B (21700279)