Mobile articles tend to be very long as the content is often shoe-horned from the web to mobile in an unnatural way. Results show readership of these mobile articles being in the single digits. This proves to be a major problem for digital media publications who are looking to optimize their content for mobile and increase consumption among users.

What it does

The app runs through the text of the article and parses a summary based on the most relevant sentences. Moving a slider you can choose whether you'd like to read 100%, 75%, 50% and 25% of the original article. The time it would take to read the article at any of these stages is also displayed.

How we built it

We began to explore the StanfordCoreNLP for word relevance analysis and build on top of it. Based on this, we were able to determine the most relevant sentences for several percentages of the article. Using the American average reading time, we were also able to estimate the time taken to read an article at any of these percentages.

Challenges we ran into

Initially we had a team member experienced in data science however, she ended up leaving the team early in the hackathon. Aside from this, working with the optimal reading percentages for every article and coding the entire app in Swift proved to be time consuming.

Accomplishments that we're proud of

We were able to come up with a functioning app that saves the user's time and lets them dictate how much time they would like to spend reading articles on mobile. Giving the user the power proves to be vital in getting the article's message to the user in the most concise way.

What we learned

We learned a lot building natively in Swift. Aside from that, the learning curve working with NLP taught us a lot.

What's next for Kompact

Future work includes working with text-to-speech APIs to make consuming articles on the go more time efficient. Additionally, we'd like to personalize the app to the user by keeping track of their reading habits and aggregating how much time they save usually.

Built With

Share this project: