Online university sucks.
Online lectures are even worse.
Online lectures lack engagement and the collaborative nature of classroom learning.
Our mission: connect students studying online
Lecturely allows students to create a room, share it, and study together.
Within a room, Lecturely provides synchronous video playing (with support for pauses, rewinds, skips) and chat functionality.
From room data, Lecturely provides unparalleled AI-powered analytics on the lecture recording for the difficult sections.
- Students will be able to see this and discuss with their lecture peers accordingly.
- Teachers will be able to see this and improve their future lectures and content.
How we built it
Our front-end was written in React and designed using Figma.
Our back-end was written using Express to manage routes between the front-end and back-end.
Our ML model was written in Python and looked at many factors of a student's lecture experience such as pauses, rewinds and fast-forwards to determine which parts of the lecture are difficult. Our regression model predicts the level of difficulty based on student behavioural features tracked through the video player. The model is overlaid against a less weighted heuristic model to initially estimate the response variable (difficulty level)
Challenges we ran into
- As the ML model was an unsupervised model, we had to figure out the best algorithms to implement to return a level of difficulty. As we needed a numerical level, we couldn’t use traditional unsupervised methods such as clustering.
Accomplishments that we're proud of
Integrating all parts of our stack. Designing a clean UI. Eating Guzman at Monash and staying overnight.
What's next for Lecturely
- Develop and launch MVP
- Secure partnership with Monash and Unimelb
- Secure seed capital / grants for opex
- User interviews, product refinement, APAC expansion
- Global university, MOOC and online ed expansion