Unconscious bias is becoming more prevalent as we are constantly being bombarded by news feeds and articles that are geared towards our preferences. It is critical that we are self-aware of our own preconceived opinions as we browse through a variety of different articles.
What it does
SpecZo is a google chrome extension which categorizes the articles and displays your bias on a scale from 0-100 for each subtopic. In addition, it quantifies the number of bias articles read and maps this data on a chart. This gives the users a visualization of their perception on the topics.
How we built it
Google Cloud Natural Language API to classify categories and to utilize sentiment analysis for detecting bias within articles. Reddit API used to retrieve article links from Reddit posts.
Challenges we ran into
- Google chrome extension restrictions on securities
- Filtering through messy and incomplete data to train the machine learning API
- Integrating the APIs into the project
Accomplishments that we're proud of
We were able to embed several APIs and integrate all of them to make a successful application.
What we learned
We do not have to build things from scratch because there are so many APIs that achieve specific tasks and when implemented together in a project, it produces an efficient solution in a relatively short time frame.
What's next for SpecZo
- Expand the database by incorporating other news article APIs to retrieve articles from their application increasing the accuracy of the machine learning model
- Detecting biases and credibility on news feed (before users click the article) in order to warn them before reading