My final goal was to create a food map to show how similar dishes and methods of cooking are similar across different regions and countries. My goal was using food to connect people in a time of division. So I planned to run NER (named entity recognition) on Wikipedia pages to get the different foods and ingredients then draw connections between other foods depending on if they were directly referenced or had similar ingredients. However, just getting Wikipedia pages under food categories took most of the time for this hackathon. So I decided to make a simple tool to simplify the process of recursively getting Wikipedia data for future projects.

What it does

What it does is basically serve as a layer over the petscan Wikipedia utility to allow users to gain information without having to scrape Wikipedia directly or download the large dump. It returns information as a son which can be easily recursed through in other projects.

How I built it

I pretty much faked an API for the Wikipedia petscan utility which I call from a Vue app served with a flask on a google cloud host.

Challenges I ran into

Figuring out how to use google cloud was more time-consuming than I expected even for hosting.

Accomplishments that I'm proud of

Nothing really since I didn't really complete my initial goal

What I learned

getting data for NER is time-consuming

What's next for Wiki-cat

Using it for the larger food map project and maybe adding the function to directly download a zip of all referenced articles.

Share this project: