Messaging help | Devpost

Inspiration

Recent research suggests that many people find their conversations don't end when the participants want them too. This could be due to either their conversations being too short, or too long. This can impact relationships both in the workplace and on a personal level.

What it does

This project aims to help our userbase best understand where the flow of their conversations are going and what the next best message is to send. Users can upload raw HTML messages and our NLP model behind the scenes will analyze the messages, providing a predictive outcome of the next few messages.

How we built it

We developed a web scraping Python script to extract messaging data from Facebook Messenger. This information is fed into a pretrained model using NLP to provide a predictive message based on the previous messages sent and messaging style. Users can extract their messages and upload them to our site and within seconds a predictive text will be provided.

Challenges we ran into

One of the challenges we ran into was working to develop a web scraping Python script. Much of the exported HTML messaging data from Messenger was extremely difficult to process, thus our script must be extremely precise on what data to keep.

Accomplishments that we're proud of

This is the first Hackathon for many of our group members, which gave us experience working with APIs through our NLP model. Additionally, we were able to create custom scripts with Regex in order to get the proper data necessary to feed into our model.

What we learned

We learned the process of making Python scripts and how to deploy our web application. This was a previously unfamiliar territory and we took on the massive challenge of scraping message data from Messenger.

What's next

We hope to add in future iterations the ability for users to analyze their messages from a wide range of messaging apps. This can include Discord, Instagram, and Whatsapp. Additionally, images and text could be further analyzed to check for overall sentiment during the conversation flow.