Case study 07 · Education
From word problem to solvable equation.
An AI tutor that translates primary-school math word problems into equations, with conversational support for students.
- Sector
- Education
- Geography
- Singapore
- Coverage
- Primary-school math
- Interaction
- Conversational tutoring
01 / The challenge
Make the word problem understandable.
A student may know how to calculate an answer and still struggle to turn a written problem into an equation. Problem2Math focused on that step: identifying the quantities, relationships, and unknown in a primary-school math question.
02 / The approach
Translate the problem, then support the conversation.
A natural-language parser extracted the structure of the word problem. An equation composer turned those relationships into a mathematical expression the student could work with.
Conversational support built with Rasa let students ask follow-up questions. PyTorch and NLP components supported the translation from written language to the underlying math.
03 / The outcome
Reported use by 12,500 students.
Problem2Math was reported to serve 12,500 students in Singapore, with primary-school word problems as its initial focus.
Students could move from a word problem to an equation and continue the conversation in the same learning tool.
What we built
The system, in parts.
- 01
Natural-language parser
Reads a word problem and extracts quantities, relationships, and the unknown.
- 02
Equation composer
Assembles the solvable equation that the parsed problem implies.
- 03
Conversational tutor
Rasa-driven dialogue so a stuck student can ask and adjust, not just submit.
Technology used
- Rasa
- NLP
- PyTorch
- Python
Selected work delivered across GOT AI, Rocket Farm Studios, and Full Cortex in engineering, technical leadership, and advisory roles. Results are as reported by the project.
Start with a conversation
Have a workflow like this?
Let’s look at it together.
Bring the process that takes up your team’s time. We’ll look at what it costs today, and whether AI, conventional automation, or a simpler fix is the right next step.