7 Second News

Image of the Front Page of the news app

Inspiration

We wanted to build a news app that would deliver informative, unbiased news in a way that is easy for people to understand. One of the members in our group has a reading disability, so we also wanted to focus on accessibility and make the app accessible for people with vision impairments and reading disabilities.

What it does

Our news app scrapes the most recent articles from various sources, uses topic modeling to group similar articles (such as an article from CNN and an article from Fox News about a Senate hearing), and summarizes the similar articles into short, easily digestible content. This way, the major points from all of the articles are included in our summarization, without any bias.

Accessibility Focus

It was found that many news apps do not have features that cater to people with vision impairments and reading disabilities such as dyslexia. Our app has accessibility features such as text-to-speech and the ability to increase the font size to help people with vision impairments. We also made a special contrast and text spacing mode to help people with reading disabilities use our app.

How we built it

We built it using react for the app design and python with flask for the backend. Topic modeling was done using a Latent Dirichlet Allocation (LDA) algorithm that was sourced from the gensim package. Text summarization was done using the openai API.

Challenges we ran into

We ran into several challenges with the topic modeling and were not able to figure out a way to tell the LDA model how many topics to split the articles into. We ended up discovering the coherence metric in the gensim package which helped us use the data to make an informed decision about how many topics to include.

Accomplishments that we're proud of

We are proud of the fact that we were able to build a functioning app within 24 hours. We didn't have a lot of experience with natural language processing prior to this so it was a great learning experience.

What we learned

We learned a lot about natural language processing and NLP model evaluation. We also learned a lot of techniques about how to make an app more accessible to people with vision impairments.