The 2016 presidential election campaign has been exciting with major candidate movements in polls on both sides of the party lines. The political system still relies on the use of manual polling. The question remains: can Web data be used as an alternative to understanding public opinion online?
In this webinar on October 21, 2015, Data Acquisition Engineer, Jamie Martin, and Rosoka’s Computational Linguist, Dr. Kelly Enochson, covered how harvesting public news media data and social media data can lead to better insights into understanding public opinion online. They explored both Twitter and news in different regions across the United States to understand the public and media opinions of candidates.
With the Internet growing at exponential rates, data harvesting is becoming a huge task. BrightPlanet can harvest data in many ways, but during the webinar, Jamie focused on data that could be collected from sources that can be used to help determine public opinion online. The main sources targeted for harvesting by BrightPlanet were news sources and publicly available Tweets.
While BrightPlanet harvests the data, Rosoka is responsible for enriching and structuring that data. Rosoka can extract names of candidates and structure them in a useful way. Through this, mentions of candidates can be rated as positive or negative towards a candidate’s image. The enrichment and structuring of the data provides a better understanding of the online public opinion of candidates.
Visualization and Analysis
Through data harvesting and enrichment, useful information about public opinion data is created. To make a visual representation of this data, Tableau is used. The visual graphs display information about the candidates and how often they are mentioned in the news and on Twitter. These graphs can compare two candidates against each other to help us further understand the online public opinion of candidates.