dckx is the news aggregator of the future! Tired of boring world news? Can't get enough programmer humor? Read your Times headlines in XKCD panels!
What it does
Creates a unique comic xkcd panel based off of the most recent The New York Times' headline.
How I built it
In order to run, dckx has a large precollated library of source material. The materials consists of webscraped XKCD graphics, caption text, and executive summaries drawn from explainxkcd.com. Input is tokenized by dckx.py, then fed into search.py, which uses several techniques to match each token to a proper comic panel. These panels are recombined in an output comic, which is displayed for the user's convenience. The panels were spliced using OpenCV and contour recognition.
Challenges I ran into
Setting up OpenCV with Python
Accomplishments that I'm proud of
What I learned
What's next for dckx
Improve Natural Language Classifiers to more accurately analyze text and create better matches between xkcd transcripts and news article headlines.