Large-Scale Network Analytics for Online Social Brand Advertising

This paper proposes an audience selection framework for on-line brand advertising based on a large amount of user historical activities on social media platforms.

It is one of the first studies to our knowledge that analyze implicit brand-to-brand networks for online brand advertising. This paper makes several important contributions:

We build and analyze two weighted networks representing interactions among users and brands. To investigate network properties, we propose a technique for normalizing relationship weights in these networks from local and global perspectives. We then explore the structure of these networks to understand their implications and propose a comprehensive framework for defining target audiences for brand advertising.

As a part of this framework, we develop a hierarchical community detection algorithm for weighted undirected networks to identify a set of target brands. We then develop influential brand identification algorithm to find important brands from this set of target brands and then use sentiment analysis to identify target users from these brands. Since our datasets and network are very large, we implement several MapReduce-based techniques under Hadoop environment, for network generation and for the network analysis algorithms. Finally, we design a novel evaluation technique to test the effectiveness of our targeting framework and algorithms. The experiments conducted on Facebook data show that our framework can obtain performance improvement in terms of the number of users who are potentially interested in the focal brand, which can be up to 250 times higher than baselines.


 

Social media graphic courtesy Shutterstock.