Through PolitiTrend, it is now possible to determine the opinions of a news source regarding a pressing societal issue over time. Spanish liberal philosopher José Ortega y Gasset wrote “Yo soy yo y mi microbioma”— I am me and my micro biome. We are the products of the environments that we are raised in. The average issue of The New York Times contains more information than an average man in the Elizabethan era would have read in his entire lifetime. With the understanding that the media is part of our “microbiome”, we feel that the media we consume (in such large quantities) can define who we are. It is vital that we understand how our media’s views on how human rights, war, and other social issues have fluctuated. As three students passionate about intersections between the liberal arts, the humanities, and engineering, we felt empowered to bridge this gap.
PolitiTrend allows users - whether they are political science scholars, sociologists, historians, or curious citizens - to examine past media slants around any issue. This is especially important because an aggregate survey of all news sources may not necessarily conclude that the political spectrum is appropriately represented. For example, 11 of the 12 capital daily newspapers in Australia are run by the media behemoth “News Limited”. Can they be trusted to appropriately represent the political spectrum? Through this data, we can understand the ideological influences on societal behaviours on a whole new level. Did aggregate media reporting bias affect the outcome of the last US Presidential Election? How has media bias regarding the war on terror changed in the years since September 11th? Answers to these questions are now possible.
How it works
PolitiTrend currently works with the New York Times. Users simply select a beginning date and an end date as well as a topic of interest in the form of a string. The website then pulls articles from the New York Times and fetches political sentiment from indico.io’s service. This is then displayed for the user in the form of stacked bar graphs to illustrate mindshare over time.
Challenges We ran into
- Indico.io’s API was struggling to handle the workload of the sheer amount of data that we were rapidly feeding it. We believe that the combined workload of a large volume of requests from HackPrinceton and UMass effectively brought the service down for the duration of the weekend.
- Google App Engine didn’t play nicely with Indico.Io’s Python API which required a lot of manual tweaking that took as long as four hours and over 20 stacktraces.
What I learned
Apart from the technical skills gained from overcoming the problems mentioned above, we were able to draw a variety of conclusions with societal ramifications.
We were able to draw several conclusions about the nature of the New York Times. The first being that there tends to a clear cyclical period in its news biases over the course of a year, interestingly enough, fluctuating between libertarianism and liberalism. This trend appeared to be present for both articles about climate change and articles about the war on terror.
What’s next for PolitiTrends
- Awaiting the resurgence of the Indico.Io API like a phoenix from the ashes of a dual-hackathon DDoS so it starts working again.
- We’re currently searching for a way to generically download news articles from different publications. The NYT was chosen as a publication because of its readily available developer tools. We’d like to analyze publications like Fox News, The Economist, The Guardian, etc.