Degree Type

Creative Component

Semester of Graduation

Fall 2017

Department

Computer Science

First Major Professor

Adisak Sukul

Second Major Professor

Wallapak Tavanapong

Degree(s)

Master of Science (MS)

Major(s)

Computer Science

Abstract

Investment in online political ad marketing is gaining traction very rapidly. In the United States, the 2016 presidential election campaign witnessed a substantial increase in political advertisement expenditure on online platforms like YouTube. Therefore, political researchers are interested in analyzing trends of political ads in an online medium. But currently, there is no existing method or application that can classify political advertisement from a large dataset of online ads. In this paper, we attempted to solve this problem by proposing a model that can automatically classify political video advertisements using machine learning algorithms such as Support Vector Machine, Linear Regression, and Naïve Bayes classifier. We will also focus on feature engineering for this classification problem. We applied text features and non-text features like color and facial features for classification purposes. We trained 3 different models with a different feature sets and compare results among them. We also created an ensemble with these 3 models and achieved an F1-score of 0.97.

Copyright Owner

Boudhayan Banerjee

File Format

PDF

Share

COinS