Deciding who to work for is one of the biggest decisions you make. Financial data, stock prices, and cold numbers just aren't enough to pin the rest of your life on.
What it does
FluxDuck analyzes companies' reputations in the news media to help you find an employer worth working for.
How I built it
The Indeed API was used to return a basic list of jobs for a given position such as "software engineer".
The app scrapes information about companies from news media sites like CNN Money using a from-scratch Python crawler. Over 800 companies were ranked by analyzing over 60,000 news articles from the past 5 years.This information is analyzed using two forms of sentiment analysis: Wolfram Cloud API sentiment analysis and a from-scratch Python sentiment analysis based on Bill McDonald's (University of Notre Dame) master dictionaries for financial sentiment analysis.
The sentiment analysis is used to rank the companies to help you decide which employers you would actually like to work for. A numerical score is returned for each company and you can view how a company's media reputation has changed over time.
Challenges I ran into
The biggest challenge is designing the algorithm to interpret the sentiment analysis data in a meaningful way. This is proprietary and will not be revealed.
Getting the web app and servers to work right is always tough.
What I learned
-Wolfram is incredible. Seriously.
-Sentiment Analysis is a really powerful tool, but you have to get your dictionaries and algorithms well tuned.