SoMin.ai: Social Multimedia Influencer Discovery Marketplace
Aleksandr Farseev
ITMO Univerity
St. Petersburg, Russia
farseev@gmail.com
Kirill Lepikhin
Social Miners Research
Singapore
kirill@somin.ai
Hendrik Schwartz
Social Miners Research
Singapore
hendrik@somin.ai
Eu Khoon Ang
Social Miners
Singapore
eukhoon@somin.ai
Kenny Powar
Rebel Owl
Singapore
kpowar@rebel-owl.com
ABSTRACT
In this technical demonstration, we showcase the frst ai-driven so-
cial multimedia infuencer discovery marketplace, called SoMin [4].
The platform combines advanced data analytics and behavioral sci-
ence to help marketers fnd, understand their audience and engage
the most relevant social media micro-infuencers at a large scale.
SoMin harvests brand-specifc life social multimedia streams in a
specifed market domain, followed by rich analytics and semantic-
based infuencer search. The Individual User Profling models ex-
trapolate the key personal characteristics of the brand audience,
while the infuencer retrieval engine reveals the semantically-matching
social media infuencers to the platform users. The infuencers are
matched in terms of both their-posted content and social media
audiences, while the evaluation results demonstrate an excellent
performance of the proposed recommender framework. By lever-
aging infuencers at a large scale, marketers will be able to execute
more efective marketing campaigns of higher trust and at a lower
cost.
ACM Reference Format:
Aleksandr Farseev, Kirill Lepikhin, Hendrik Schwartz, Eu Khoon Ang,
and Kenny Powar. 2018. SoMin.ai: Social Multimedia Infuencer Discov-
ery Marketplace. In 2018 ACM Multimedia Conference (MM ’18), October
22–26, 2018, Seoul, Republic of Korea. ACM, New York, NY, USA, 3 pages.
https://doi.org/10.1145/3240508.3241387
1 INTRODUCTION
The past decade has testifed a rapid growth of the Internet. One
can observe the drastic expansion of social networking services,
where millions of users publish and consume information regu-
larly. Built upon such growth, social media marketing industry has
correspondingly developed its capabilities of helping marketers
in content personalization and deliverance [11ś13]. However, the
growing amount of irrelevant content, such as unrelated advertise-
ment and spam, made social media users more and more reluctant
towards perceiving sponsored search results and online advertise-
ment, such as łGoogle AdWordsž [10] and łFacebook Sponsored
Adsž [1].
Permission to make digital or hard copies of part or all of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for proft or commercial advantage and that copies bear this notice and the full citation
on the frst page. Copyrights for third-party components of this work must be honored.
For all other uses, contact the owner/author(s).
MM ’18, October 22–26, 2018, Seoul, Republic of Korea
© 2018 Copyright held by the owner/author(s).
ACM ISBN 978-1-4503-5665-7/18/10.
https://doi.org/10.1145/3240508.3241387
To mitigate such customer skepticism, marketers often leverage
on human-centric content delivery channels, where Infuencer Mar-
keting clearly dominates over other marketing strategies. Indeed,
it has been shown that 92% of consumers are more likely to trust
brands that advertise via infuencer channels [17] rather than those
who has adopted conventional marketing strategies. Unfortunately,
the limited availability of infuencer search platforms and the ab-
sence of audience-based and content-based infuencer matching
technology, result in tremendous amounts of manual work per-
formed, the corresponding high marketing agency service costs
and low efciency of the conducted marketing campaigns.
Aiming at bridging the aforementioned research and industrial
gaps, in this technical demonstration we propose an online infu-
encer discovery platform that would perform infuencer matching
at both content and audience levels simultaneously. To accomplish
such semantic search, we have utilized our previously-proposed
multi-source re-ranking approach [6], which is able to perform
well-balanced and source-consistent recommendation over multi-
ple data representations. The content representation was gained via
extracting multi-modal hot topics, named entities, and image con-
cepts [15] from recent infuencer-posted and brand-intended con-
tent, while the audience representation was obtained via computing
distributions of the brand and infuencer social media followers
over automatically-profled behavioral user attributes [3, 7, 8]. The
above technology was integrated into the cloud-based social mul-
timedia infuencer discovery marketplace SoMin, which delivers
semantic-based infuencer search to the corporate users.
The overall data processing pipeline in SoMin is as follows: (a) the
distributed cloud crawlers harvest recently posted user-generated
content (UGC) from multiple social networks with respect to a
particular brand and its corresponding social media infuencers [2];
(b) user profles are predicted via SoMin’s Social Multimedia Ana-
lytics API endpoints [15] for all brand’s and infuencers’ followers,
which allows for drawing brand-specifc and infuencer-specifc
bahavioural audience distributions; (c) multi-source multi-modal
topic, named entity recognition, as well as image concept detec-
tion models [15] extract textual and visual content representations,
which are used for drawing infuencers’ semantic content distri-
butions as well as the semantic description of the brand-intended
marketing message; and (d) the multi-modal re-ranking engine [6]
recommends semantically-matching social media infuencers with
respect to the brand-intended marketing message and its audience.
To the best of our knowledge, SoMin is the frst ai-driven social