Inspiration
Group chats can get extremely long – and when you return to a chat after missing several hundred messages, scrolling through the clutter to find the relevant information can be a trying task. Context Chat is designed to be a smart messaging platform designed to solve this problem. Created specifically with online group chats in mind, Context Chat allows users to create and join group chats at givecontextto.me.
What it does
The power of Context Chat lies in its summarizing functionality: for users just logging in to a group chat, or for users who may just wish to view the important points of what was said, Context Chat goes through the chat history and finds which messages are important – these messages are then conveniently displayed in a sidebar for the user.
How we built it
The actual mechanics of determining which messages to show rely on a heuristic scoring system based on results from Google Cloud Platform’s Natural Language Processing API. The system accounts and weights several factors, including sentence complexity, mention of entities, proximity to earlier questions, and elapsed time.
Challenges we ran into
Throughout the event, we were faced with two main problems. The first was incorporating our different ideas and tasks given the very little time limit that we had. The second problem was synchronizing our working environments together. Hack the North was the first time we all met each other so we had little awareness of each others capabilities, weaknesses and strengths. As the each of us worked on different components of the project, there was various delays when trying to connect our components together. For an example, setting up the database took longer than expected and the back-end side of the program could not continue working without getting sample data sets to test the logic. However, despite these challenges, we all held a positive mindset and team dynamics was positive at all times ensuring we could all work to the best of our abilities.
Accomplishments that we're proud of
We are very proud of being able to manage to incorporate the different sections of our work together completing our core goals in the time limit that we had. Our goal at the start of the Hackathon was to learn new skills and take at least one skill away from the event. We decided to work on APIs that we have never used before(Google Cloud API and Cockroach) to accomplish these goals and challenge ourselves. We are proud to say that all four us have definitely taken away more than just one skill from the program. We all worked in areas that capitalized on our strengths while also touching onto other group members works to gain a better understanding of the different components of our program.
What we learned
We have learned to trust each others capabilities and not shy away from asking questions when stuck solving a difficult question. Overall we have walked away from the event with not only better technical skills but also improved team-working a different mindset.
Whats next for Context Chat
Future work on Context Chat would focus on refining the message selection algorithm – most likely using machine learning on a set of training documents to determine the ideal weights for the different factors. Additional features would also focus on making the platform more ‘smart’ – for example, using the Yelp API in messages to link to mentioned businesses.

Log in or sign up for Devpost to join the conversation.