A tool that auto generates questions in QuML format in English language

[sajeshkayyath]

Project Details

What is Sunbird inQuiry?

Sunbird inQuiry is a software building block that enables setting up of question banks for various use cases such as assessments, quizzes, practice worksheets, surveys, and many more. These are applicable in multiple domains related to education and human development.

Here are some examples of solutions that can be enabled through Sunbird inQuiry:

• Conduct assessments which can evaluate users’ understanding of concepts and issue certificates based on assessment score.
• Evaluate assessment data to take informed decisions and plan for interventions.
• Launch surveys to collect feedback and analyse responses.

Question Set Editor (QuML editor)

Question set editor is a tool offered by inQuiry to help you kickstart your creation of Question Banks.

More details are here

Features to be implemented

A tool that auto generates questions in QuML format in English by taking a story/chapter as an input - using AI

The questions are currently prepared manually and injected using the QuML editor. As part of this requirement, a standalone tool has to be build that can take in a story/chapter as input in English and auto generate questions in QuML format.

Questions that are auto generated will be of type multi choice and subjective.

What exists

• QuML Specification
• QuML player that is capable of playing multi choice and subjective questions.

What needs to be build

• A standalone AI tool that can take in a story/chapter as input and auto generate questions in QuML format
• Schema validation to ensure the generated question complies to QuML format.
• Questions that are auto generated are of type multi choice and subjective questions

Learning Path

Complexity
Medium

Skills Required
Open source AI tooling
JSON Schema

Name of Mentors
@krgauraw

Project size
6 Weeks

Product Set Up

Details of the QuML specification is available here. The software stack that is needed for the standalone application is to be setup newly.

Acceptance Criteria

• New tool is able to take input as story/chapter and auto generate questions
• Questions generated are complying to QuML Specification
• Open source software to be used
• No vendor lock in

Out of Scope

Enhancing existing QuML player and QuML Editor
Implementing additional question types apart from multi choice and subjective

Milestones

• Understanding the requirements
• Understanding QuML Specification
• Submit the solution for the new tool
• Solution approval
• Setting up the app
• Generating questions from any given story/chapter
• Generating question complying to QuML specification
• Validating the questions against the QuML specification
• Testing the application to verify different question types
• Good to have: Solution proposal for auto generating questions for new types like Fill in the blanks and match the following.

[prakharrathi25]

Hi, would we be building our own AI solutions in this or will it require pre-existing LLM capabilities. I have worked on both in the past and would love to work on this project!

[prashant2-4-4]

Hi , I would want to work on this. from resource point of view If resources are limited, building your own AI system may be a viable option. Open-source LLMs like Llama 7B require significant GPU resources (14GB+) for inference. However, if sufficient resources are available, utilizing pre-trained LLMs can be beneficial since they are trained on extensive datasets.

[goyalpramod]

Heya there @sajeshkayyath my understanding of the product and the project is. Inquiry provides a platform in which you can input text in the question markup language. And it generates questions for you.

I have but three questions.

In the overview video it has been mentioned that inquiry is capable of authorising and publishing. Using an external API.

1. What is the input that is being provided to this API?

2. Do we have access to this API?

3. The AI tool that we are to develop. What is expected of us? Because we certainly do not have the resources (as an individual) to train an LLM

My further understanding is, after developing the AI tool. We need to develop an interface that can take in a story and develop the questions in QuML format.

(Kindly let me know if i got anything wrong)

[vaibhavbhuva]

Hi All,
We are conducting a 30-minute Q&A and issue clarification session to address any doubts regarding proposal submission or any other concerns to clarify.

Here are the session details for the connect:

Date: 9th June 2023 - Friday (Today)
Time: 05:30 PM to 06:00 PM
Meeting Link: https://teams.microsoft.com/l/meetup-join/

Kindly join discord channel for continuous updates.
Looking forward to meeting you all.

Thank you.

[krgauraw]

Hi @prakharrathi25 / @prashant2-4-4 / @goyalpramod Are you guys writing proposals? If yes, please submit it on unstop platform.

[goyalpramod]

@krgauraw heya there, I have submitted a proposal in unstop. Kindly let me know if it's not visible. Because it shows submitted from my side.

[prashant2-4-4]

Yes, I am writing and I will submit it by today.

…

On Sat, 10 Jun 2023 at 00:24, Pramod Goyal ***@***.***> wrote: @krgauraw <https://github.com/krgauraw> heya there, I have submitted a proposal in unstop. Kindly let me know if it's not visible. Because it shows submitted from my side. — Reply to this email directly, view it on GitHub <#41 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ANKXDYQ2AGCJGJMR6QGYOGDXKNWPJANCNFSM6AAAAAAXOKB5FI> . You are receiving this because you were mentioned.Message ID: ***@***.***>