Abstract:
Abstract—Influence on the social network platform has become
an interesting research area. A typical social network influencer
is known by the amount of reactions that he/she attracts based
on the posts made. The amount of reactions received is usually
a reflection of the overall visibility of the influencer on the social
network arena. This visibility has attracted commercial benefits
to such influencers through word of mouth marketing, political
endorsements as well as ambassadorial appointments. However,
existing methods in literature that are used to scientifically
quantify the amount of social influence that can be attributed
to an influencer, tend to lump the amount of social influence in a
single basket labelled either as positive or negative influence. The
effect is a binary classification of influencers as either negative
or positive. In this article, we make the case that this is not
necessarily accurate. The reason being that in a collection of
comments that a post from an influencer attracts, it it seldom that
all the comments would be absolutely positive or negative. There
is a fuzzy space defined by comments that do not necessarily
belong to either of these categories.
In this work, we have used data harvested from Facebook
comments and grouped into four different categories - those in
full support of the opinion expressed, those in full opposition
of the opinion, those that somewhat support and those that
somewhat oppose the opinion expressed. Results have shown
that the fuzzy space created by the comments in between the
full support and full opposition negates the assumption that an
influencer can be a positive or negative influencer just because
the number of positive or negative comments are the majority.
Index Terms—Social Influence, Influencer, Fuzzy Space, Multiclass Classifier, Social Network