Abstract
Individuals inadvertently allow emotions to drive their rational thoughts to predetermined conclusions regarding political partiality issues. Being well-informed about the subject in question mitigates emotions' influence on humans' cognitive reasoning, but it does not eliminate bias. By nature, humans tend to pick a side based on their beliefs, personal interests, and principles. Hence, journalists' political leaning is defining factor in the rise of the polarity of political news coverage. Political bias studies usually align subjects or controversial topics of the news coverage to a particular ideology. However, politicians as private citizens or public officials are also consistently in the media spotlight throughout their careers. Detecting political polarity in the news coverage of politicians rather than topics adds a new perspective. Determining the best approach for detecting political polarity in the news relies on the news delivery method. Data types such as videos, audio, or text could summarize the news delivery methods. Text is one of the most prominent news delivery methods. Text pattern recognition and text classification are well-established research areas with applications in many multidisciplinary domains. We propose to use deep neural networks to detect ideology in news media articles that cover news related to political officials, namely, President Obama and Trump. Deep network models were able to identify the political ideology of articles with over 0.9 F1-Score. An evaluation and analysis of deep neural network performance in detecting political ideology of news articles, articles' authors, and news sources are presented in the paper. Furthermore, this paper experiments on and provides a detailed analysis of newly reconstructed datasets.