Credit Distribution for Influence Maximization in
Online Social Networks with Time Constraint
Yan Pan
∗
, Xiaoheng Deng
†
, Hailan Shen
‡
School of Information Science and Engineering
Central South University, Changsha, China, 410083
Email:
∗
panyan@csu.edu.cn,
†
dxh@csu.edu.cn,
‡
hailansh@mail.csu.edu.cn
Abstract—Considering the time constraint, influence maxi-
mization with time constraint (IMTC) is a problem of identifying
several maximum influential individuals as seed nodes who will
influence others and lead to the largest number of adoption
in an expected sense. Associated with probabilities of events
and the radio of information gain, we propose an optimized
approach to evaluate the activation probability synthetically.
As the credit which indicates the strength of influence given
to adjacent neighbors is depended on the optimized activation
probability (OAP), we also extend the Credit Distribution (CD)
model by restricting the scope of credit distribution with the
time-delay aspect of influence diffusion in online social networks.
Furthermore, the time obstacle caused by repeated attempts
is converted to length of the action propagation augmented
paths (APAP). The simulations and experiments implemented
on real datasets manifest that our approach is more effectively
and efficiently in identifying seed nodes and predicting influence
diffusion compared with other related approaches.
Keywords—online social networks; influence maximization;
credit distribution; time constraint; greedy algorithm
I. INTRODUCTION
With the development and popularity of social platform
services, such as Facebook, Twitter, and WeChat, the study of
influence maximization in online social networks has attracted
a great deal of attention in recent years. Without distance
limitation, the information can be propagated extremely widely
and rapidly. At the same time, to boost the benefits from viral
marketing, existed social network services usually take advan-
tage of influence diffusion in online social networks to amplify
the prestige of their brands such that the adoptions of total
products are as large as possible. However, one of fundamental
challenges for viral marketing is that how to find a part of
individuals to maximize the word-of-mouth propagation in a
certain period of time. Conventionally, we make this problem
as Influence Maximization with Time Constraint(IMTC), i.e.,
with some time restricts, how to find a set of initial individuals
such that obtaining the maximum expected number of users
influenced after propagation in social networks. Intuitively,
this problem is quite practical because the constraints are
always objective existed in most real circumstances. Most
of marketers feel more satisfied to pursuit maximum profits
under minimum investments, rather than identifying arbitrary
users and wasting resources to achieve a blind propagation
result by luck. Meanwhile, it is closely associated with reliable
prediction of influence diffusion and quality of seed nodes,
because we need to predict how likely the users will be
influenced by the initial ones.
In contrast to most traditional media, influence spread in
online social networks is mainly based on the fictitious rela-
tionships such as credit and affinity. Because of lacking region
restrictions, information can be propagated at extremely high
speed with exceedingly low cost. However, based on factors
such as similar preferences and characters, the relationships
among users are fragile and volatile. Consider the following
hypothetical scenario as an motivating example. A company
designs a new product and wants to market it through an online
social network, but owning to the limited budget, it could
only select a small set of initial users to persuade. At the
same time, the managers in company always want to enlarge
their product adoption in a short time frame, in hoping that
through the word-of-mouth effects, a large number of product
adoption will occur [13]. Beyond that, they always make time
schedule and predict the influence spread so that they could
adjust their strategy simultaneously, which means we should
do fully incorporate the temporal aspect that has been well
observed in the dynamics of influence diffusion.
Specifically, the propagation of influence from one to anoth-
er may incur an uncertain period of delayed time. For example,
instead of presenting in one common place or having a face-
to-face conversion, the influence activities are indicated by
sending and receiving messages in online social networks. But
in practical scenario, one can not always stay online, even
though disregarding the information be seen by the receiver
is a random event which occurs occasionally, there still exists
a short period of time between that he opens his terminal
and links to the services. So that the time-delay aspect is
essential in the process of influence diffusion. On the other
hand, both the strength and spread of influence are time-
sensitive. In a certain viral marketing campaign, just like the
hot topic vanishes with time elapsing, the influence about
the messages will be weaken, and the spread of influence
will be alleviated. This is the key reason that the managers
of companies wish to trigger a large volume of product
adoptions in a fairly short time period. Therefore, if we try to
maximize the influence spread for a viral marketing campaign
facing practical scenarios, we need to take temporal effect of
influence diffusion into consideration.
2015 IEEE International Conference on Smart City/SocialCom/SustainCom together with DataCom 2015 and SC2 2015
978-1-5090-1893-2/15 $31.00 © 2015 IEEE
DOI 10.1109/SmartCity.2015.80
255