Inspiration

Students in Singapore spend significant amounts of time doing practice questions, which takes a toll on them. We at Revise Smart have experienced that.

Typically, students simply try to do all questions. That is ineffective -- even within a specific topic, students are strong at certain kinds of questions, and weak at other kinds.They should focus on what they're weak at, but it's hard to do that using assessment books.

We want to empower students to revise in a smarter way, allowing them to do more for less, thereby having more free time.

What it does

A machine learning powered online assessment platform for students.

Based on students' responses to questions, the platform identifies topics they are weak at, and provides a tailored set of questions targeted at their weaknesses.

This allows students to study more effectively, potentially saving hours every week.

Furthermore, unlike traditional assessment books, which only provide workings for questions, the platform will provide explanations for workings (generated by algorithms), which will better help students understand how to solve the problems.

How we built it

1) We built a model to automatically group questions into different topics. We scraped the internet for Math questions, then used the Latent Dirichlet Allocation model to group questions based on keywords determined by the model. Several types of preprocessing were done before-hand.

2) To automatically generate explanations for working, we used natural language processing (NLP) techniques to extract keywords from the questions, and matched these keywords with the workings to determine what the purpose of each step was.

Challenges we ran into

It was hard to find suitable Math question banks online. NLP techniques can be pretty complicated. Website formatting can be a huge pain.

Accomplishments that we're proud of

We managed to build a few models we're unfamiliar with overnight, with pretty good results.

What we learned

NLP techniques and models. Front-end web frameworks.

What's next for Revise Smart

If we think it's promising, perhaps we'll continue building on what we have, bringing it live, and helping the 450 thousand students in Singapore (and many more overseas) revise more efficiently!

Share this project:
×

Updates