Date of Award

8-2014

Document Type

Thesis

Degree Name

Master of Science (MS)

Legacy Department

Mechanical Engineering

Committee Chair/Advisor

Wang, Yue

Committee Member

Wagner , John R.

Committee Member

Shen , Haiying

Abstract

Internet has undoubtedly become a major and effective medium for advertising, and it will soon replace traditional advertising media. Online advertising ecosystem is consisted of thousands of agents, which can be divided into three types: users, advertisers and publishers. Each type of the agents plays an important role in the ecosystem. However, it is generally very difficult to model such a complex and large-scale system, not to mention the design of suitable strategies for the ecosystem. On the other side, replicator-mutator (RM) dynamics combine tools from evolutionary dynamics of populations, complex networks, and control theory to study evolution of behavior of agents in social networks. RM dynamics have been widely used in multi-agent networks to mimic an agent's evolutionary states based on its own decision as well as environmental effects. This thesis develops an analytical framework of a naturally fitted coevolving online advertising ecosystem based on RM dynamics in order to capture the evolution of all participants and design suitable advertisement strategies. Compared with real biological systems, online advertisement ecosystems share a lot of similarities. These similarities not only affect participants' local characteristics, but also contribute to performance of the entire ecosystems. We first introduce the RM dynamic model for the user agents and advertiser agents, respectively. The effects of major parameters in the RM dynamics are studied. We demonstrate the evolution of the replicator dynamics, mutator dynamics, and RM dynamics using a toy problem. Next, we provide a detailed example using data crawled from a real online advertising ecosystem for tablet sales. We evaluate each parameter in the advertiser and user RM dynamics using the crawled data. We show the changes in ranking places and sales volumes of four chosen advertisers (Apple, Asus, Sumsung and Kindle) at four different stages based on the chosen parameters and corresponding evolutionary dynamics. Finally, we discuss suitable strategies for online advertising based on the proposed framework and conclude the thesis with a summary and discussion of future work.

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.