Headline generation is an area of active research in Computer Science x Journalism. We decided to try tackling it ourselves with some of Microsoft's APIs!

What it does

Headlines is a language processing tool that reads news articles and outputs a headline for the article.

How we built it

It uses Microsoft Text and Linguistic Analysis to process the text into filtered and scored keyword clusters.

Challenge we ran into

Right after submitting, Microsoft's Linguistic Analysis API went down. We talked to the sponsers and they confirmed that it was currently not working. Fortunately, Microsoft's Key Words API was still up and running, so we could still keep the core feature of our app, but we threw out a lot of grammatical processing and cleaning.

Accomplishments that we're proud of

Our interface looks nice and clean, and our headlines successfully capture the essense of the articles!

What we learned

Plan ahead! A significant amount of work went unused during this project due to either incompatibilities or change of plans.

What's next

We're really really looking forward to tackling this project with a neural network. After our experience using key-word and linguistic analysis, we're curious how an NN would fare.

Built With

  • heroku
  • microsoft-linguistic-analysis
  • microsoft-text-analysis
Share this project: