Inspiration

We live in the city, and it's useful to get up-to-date information about what is happening around town. There is already a city-data forum where people can discuss events and ask questions, so we decided it would be useful to organize this information into a simple, browsable, and categorized front page.

What it does

It crawls along the city-data forum, passing all of the links to IBM's Alchemy Language parser. This pulled the text out of the HTML of the forum posts. We then passed this into the IBM Watson Natural Language classifier, which we hand trained on the first two hundred posts to classify into one of several categories. The web page (will) display each of the categories when clicked, pulling the data from a database.

Challenges we ran into

We lost our front end developer in the morning, so we had to scramble to get a visible interface. (It still doesn't exist.) We also had to swim through lots of Watson documentation because we didn't know the technology.

Accomplishments that we're proud of

The web crawler is both powerful and accomplished quite beautifully. We also are proud of figuring out how to use the Watson Natural Language Classifier to classify such a diverse set of data.

What we learned

We learned how to make web crawlers, access databases, use industry-grade APIs for machine learning, read instructions, stack cups, and code fast.

What's next for PittForum

We will finish the website and get it displayable.

Share this project:

Updates