This is the architecture diagram.
An informed electorate is as vital as the ballot itself in facilitating a true democracy. In this day and age, it is not a lack of information but rather an excess that threatens to take power away from the people. Finding the time to research all 19 Democratic nominee hopefuls to make a truly informed decision is a challenge for most, and out of convenience, many voters tend to rely on just a handful of major media outlets as the source of truth. This monopoly on information gives mass media considerable ability to project its biases onto the public opinion. The solution to this problem presents an opportunity to utilize technology for social good.
What it does
InforME returns power to the people by leveraging Google Cloud’s Natural Language API to detect systematic biases across a large volume of articles pertinent to the 2020 Presidential Election from 8 major media sources, including ABC, CNN, Fox, Washington Post, and Associated Press. We accomplish this by scraping relevant and recent articles from a variety of online sources and using the Google Cloud NLP API to perform sentiment analysis on them. We then aggregate individual entity sentiments and statistical measures of linguistic salience in order to synthesize our data in a meaningful and convenient format for understanding and comparing the individual biases major media outlets hold towards or against each candidate.
How we built it and Challenges we ran into
One of the many challenges we faced is learning the new technology. We dedicated ourselves to learning multiple GCP technologies throughout HackMIT from calling GCP API to serverless deployment. We employed Google NLP API to make sense of our huge data set scraped from major news outlets, Firebase real-time database to log data, and finally GCP App Engine for deployments of our web apps. Coming into the hackathon with little experience with GCP, we found the learning curve to be steep yet rewarding. This immersion in GCP technology renders us a deeper understanding of how different components of GCP work together, and how much potential GCP has for contributing to social good.
Another challenge we faced is how to represent the data in a visually meaningful way. Though we were able to generate a lot of insightful technical data, we chose to represent the data in a straightforward, easy-to-understand way without losing information or precision. It’s undoubtedly challenging to find the perfect balance between technicality and aesthetics, and our front-end design tackles this task of using technology for social good in an accessible way without compromising the complexity of current politics. Just as there’s no simple solution to current social problems, there’s no perfect way to contribute to social good. Despite this, InforME is an attempt to return power to the people, providing for a more just distribution of information and better informed electorate, a gateway to a society where information is open and accessible.
What's next for InforME
Despite our progress, there is room for improvement. First, we can allow users to filter results by dates to better represent data in a more specific time range. We can also identify pressing issues or hot topics associated with each candidate via entity sentiment analysis. Moreover, with enough data, we can also build a graph of relationships between each candidates to better serve our audience.