ChatWithSamuel

Chat room
User Login
Sample Text Summarizer

Inspiration

As a student, whenever a friend sends their notes or simply just reading some articles to prepare for exams, do you ever feel stressed over the huge chunks of words and endless pages? Now what if I told you that I have a solution for you? Huge long chunks of text in files can be summarised to clean and clear cut paragraphs for our learning and allowing us to allocate our time more efficiently. Our group has decided to take on ChatWithSamuel chat room project to not only chat with our friends but also gain greater efficiency while texting.

What it does

Currently, our chat bot has several open chat rooms, where users can join in the group chat. What can we do in the chatroom? How is this different from those chat rooms already widely used in the market?

Apart from catching up with families and friends in the chatroom, users can also enjoy greater efficiency when using our chatroom. What do we mean by greater efficiency? When users want to send a file, the app can do an extractive summary so users can send a short summary rather than long text files to other users.

How we built it

For the extraction of text from files and images, we used optical character recognition for images and PDF, DOC, TXT reader for files in python.

For natural language processing, We used a BERT-base model in order to do extractive summarization.

For our chatroom, we used nodeJS and expressJS to build our chat interface and functionality.

Challenges we ran into

As it was our first hackathon, we had challenges building the frontend and connecting it to the NLP functions as well as making our application go live.

We had to remove certain features from our NLP due to time constraints. In a mere 24 hour period, we were required to not only create our deep learning summarizer using natural language processing (NLP) but also ensure that our chat room is ready for people to send messages across the room. Thus, we worked on extremely tight schedules and had to remove and change certain features on the spot.

Accomplishments that we're proud of

We were proud of completing this hackathon and seeing our codes get deployed live. We have successfully created our first text summarizer and chat room which we believe is truly impactful and useful to many.

Being able to get the fundamentals of a chatroom up working as well as some deep learning NLP features done is definitely something we are all proud of and an achievement we thought was impossible within 24 hours.

What we learned

We learnt how to build a chat application and integrate some NLP functionalities.

Apart from technical skills, we also learnt the importance of perseverance during a hackathon and the importance of working together as a team in order to reach our goals.