Members (IDs)

  • Calvin Tantio (C346)
  • Ivan Andika Lie (C98)

Prizes we're opting for

  1. Most Socially Useful Hack
  2. Top 8 Teams
  3. Coreteam's Best Roll


We observe the large difference between the number of Coursera course instructors and the number of students enrolled in the course. This makes it extremely difficult for instructors to attend to every single discussion forum post. Some studies have shown that the lack of forum interaction has led to a high student turnover rate, i.e. not finishing the course. This inspires us to embark on this project.

What it does

It helps Coursera course instructors (and fellow students) to allocate their limited time and resources in the discussion forum by ranking each forum post based on its need to be intervened, i.e. replied. It is hoped that this will able to minimise the attrition rate and maximise student's learning.

The application does the following:

  1. Scrapes Coursera discussion forum posts information from a web page
  2. Sends the data scraped to a server
  3. Generates a score for each post using a pre-trained hierarchical Long Short-Term Memory model
  4. Injects the generated data back into the web page

How we built it

Our design and engineering decisions are inspired by the tools that would make the user experience absolutely seamless. We used Chrome APIs to design and implement our Chrome extension, while the server was built using Flask. Our Deep Learning model (hierarchical LSTM) was developed using PyTorch and trained on historical discussion forum data.

The use of the Chrome extension allows us to scrape and inject data directly into the web page. This makes it easier for Coursera users as they can view the score directly on every post as they browse through the discussion forums. Moreover, our server provides a RESTful API so that it can be integrated into other (related) future applications.

Challenges we ran into

It was our first time developing a Chrome extension and a web server using Flask. It was not easy integrating the backend and the frontend components of our application, with us spending all night debugging and learning the nuance of our project. Nonetheless, the whole experience was very rewarding and we are happy with the result.

Accomplishments that we're proud of

With only the two of us in the team, we were proud to be able to seamlessly integrate the different components of our project within the time constraint. Despite our inexperience with frontend development, we were satisfied with the end result.

What we learned

All of the above. The nature of the project is something that both of us have not tried before. We embarked on this project as we believe that it is something that can benefit add value to the learning process of students.

What's next for CourSeraSera

We believe that the front-end and the UI component can be enhanced. Furthermore, a better model can be devised, for instance using Transformer Network. Due to the limited time and manpower, we believe that the project could be enhanced further through more thorough data cleaning and processing. Therefore, we urge fellow enthusiasts in this domain to create their own models and test it using our in-built chrome extension.

Share this project: