-
main screen
Inspiration
We thought it would be cool to use machine learning tools on current events to get a sense of what people are thinking about a particular topic. Specifically, our interest in finance inspired us to create this tool to analyze stock tickers in real time.
What it does
This app takes in any topic or keyword, sanitizes it, and uses an api to fetch an rss feed of news related to that topic. We then use an api to convert this data to json and grab the content from the news articles. We use a Bayes model ML algorithm to determine the positive or negative sentiment of each news article, and return the average score as a percentage from -100% to 100% as completely negative or positive respectively.
How we built it
HTML5/CSS3 front-end. Back-end uses Javascript and JQuery on a serverless platform, utilizing multiple REST APIs to perform functions.
Challenges we ran into
Fetching news articles was hard because there was no direct API. We had to grab the rss and convert it to JSON ourselves. There was also no easily available REST API for machine learning, so we just implemented our own Bayes algorithm based on existing positive and negative training data available online.
Accomplishments that we're proud of
Creating an app that has a complex algorithm running in the back-end while simultaneously creating a very clean, user-friendly front-end
What we learned
Machine Learning skills
What's next for PRogativ
Improving the training model to make it more accurate. Using social media feeds as part of the public sentiment algorithm.
Log in or sign up for Devpost to join the conversation.