RAG_Hack icon indicating copy to clipboard operation
RAG_Hack copied to clipboard

Project: Interactive Learning Platform

Open lilbandit opened this issue 1 year ago • 4 comments

Project Name

Curio

Description

✨Curio

Curio is a personalised learning platform which uses Retrieval-Augmented Generation (RAG) to generate interactive audio lessons that engage users in a dynamic learning experience. The users must simply upload study materials like books or lessons and Curio would create conversational and adaptive lessons based on the users' needs.

🗃️Data Sources

Curio relies on user uploaded PDFs to generate the speech-driven learning content. These PDFs can include anything from individual chapters to complete textbooks. This material is indexed and stored it in a vector database, which serves as the knowledge base for our RAG system.

⚙️Additional modifications

To improve content extraction, Curio employs specialized parsing techniques focused on:

  • Heading and Subheading Extraction
  • Content Chunking
  • Metadata Association

These techniques focus on identifying and extracting headings, chapters, and subchapters, allowing us to generate precise chunks related to specific subsections of the material. This structured approach ensures that each chunk is tagged with relevant metadata, including its heading and subsection, making it easier to provide personalized, interactive learning experiences.

Modifications to Data ingestion Pipelines: A modified script is created by performing enriched generation of the content extracted from the file.

Modifications to prompting flow: Curio engages human in the loop interruption based agent

🧑‍🎓Beneficiaries

Curio primarily targets students from middle school to university, who can benefit from interactive, self-paced learning. Additionally, lifelong learners and professionals seeking flexible education solutions can also use the app to enhance their knowledge in a way that adapts to their busy schedules.

🌐 Architecture

The diagram below elucidates how Curio uses RAG to curate topic wise interactive audio lessons for the user. image

Technology & Languages

  • [X] JavaScript
  • [ ] Java
  • [ ] .NET
  • [X] Python
  • [X] AI Studio
  • [ ] AI Search
  • [ ] PostgreSQL
  • [X] Cosmos DB
  • [ ] Azure SQL

Project Repository URL

https://github.com/samanvithms2002/microsoft-hack

Deployed Endpoint URL

No response

Project Video

https://drive.google.com/file/d/1dXdW_NSpgP8EaxSNakP-_ucLrRli8l0V/view?usp=sharing

Team Members

chirayurathi, chaitu860, NikunjBedia, samanvithms2002, Vaishnavi-Kocherla, DedAvocado, lilbandit

lilbandit avatar Sep 17 '24 06:09 lilbandit

@lilbandit any chance you have a deployed endpoint I can test myself?

jaydestro avatar Sep 18 '24 16:09 jaydestro

@jaydestro - We haven't deployed the application publicly for now to manage costs and optimize API usage, since we are using trial versions for a few functionalities. If you could share your preferred timeline for testing, we’ll ensure the infrastructure is set up in time accordingly. Thank you!

lilbandit avatar Sep 18 '24 19:09 lilbandit

Hello @lilbandit, thank you for participating in RAG Hack!

The team is working hard to distribute badges. Please have each team member fill out this form: aka.ms/raghack/badge-dist

Our competition rules state that teams can have a maximum of 4 members, please make sure to follow the rules next time!

Thank you!

multispark avatar Oct 23 '24 00:10 multispark

@Vaishnavi-Kocherla,

Unfortunately, there have been some delays with the distribution of badges - we have been having some technical issues. Please resubmit the following form.

Form link: https://aka.ms/raghack/survey

We aim to distribute by the end of this month. Thank you for your patience!

multispark avatar Dec 05 '24 21:12 multispark