Identification of Anonymous Identical Users on more than one social media networks

Abstract

The last few years have witnessed the emergence associate degreed evolution of a vibrant analysis stream
on an outsized form of on-line Social Media Network (SMN) platforms. Recognizing anonymous, but identical users
among multiple SMNs remains degree refractory disadvantage. Clearly, cross-platform exploration might facilitate
solve many problems in social computing in every theory and applications. Since public profiles area unit usually
duplicated and easily impersonated by users with fully completely different functions, most current user identification
resolutions, that mainly concentrate on text mining of users public profiles, square measure fragile. Some studies
have tried to match users supported things and temporal property of user content additionally as style. However, the
locations square measure skinny at intervals the bulk of SMNs, associate degreed style is difficult to form out from
the short sentences of leading SMNs like S in an passing Microblog and Twitter. Moreover, since on-line SMNs
square measure quite symmetric, existing user identification schemes supported network structure don’t appear to be
effective. the $64000 world friend cycle is extraordinarily individual and nearly a pair of users share a congruent
friend cycle. Therefore, it’s further correct to use a relationship structure to analyze cross-platform SMNs. Since
identical users tend to line up partial similar relationship structures in varied SMNs, we've got an inclination to
project the Friend Relationship-Based User Identification (FRUI) algorithmic rule. FRUI calculates a match degree
for all candidates User Matched Pairs (UMPs), and alone UMPs with high ranks square measure thought of as
identical users. we've got an inclination to to boot developed a pair of propositions to spice up the efficiency of the
algorithmic rule. Results of intensive experiments demonstrate that FRUI performs much better than current network
structure-based algorithms.