We wanted to help users save time by quickly skim through massive amounts of unread chats and summarizing the most important points.

What it does

Identifies and highlights important conversations, assesses sincerity, variety, length, and quality in messenger messages. Users can click important summarized messages and be directed to the message itself. A visualization of the most important words and key themes are also shown so users can quickly conclude what they missed.

How we built it

We used python for our backend, Javascript, HTML, CSS, and web scraping to integrate our platform.

Challenges we ran into

Parsing the Facebook DOM in a chrome extension and creating an algorithm to filter for meaningful messages were some challenges we ran into. The most difficult challenge was definitely how we could compare and measure the quality of messages, and pick out what would be relevant for users.

Accomplishments that we're proud of

The filtration of terrible messages, building an intuitive interface, learning how to navigate Facebook messenger's DOM, and for some of us -- attending our first hackathon!

What we learned

Summarizing short pieces of fragmented conversations is far more difficult than summarizing large articles, generating a good summary for text data is difficult, and there are many factors in measuring the relevance and comparing messages.

What's next for Messenger Summarizer

ICO, more ML and NLP components. Add more functionality to the chrome extension for users.

Share this project: