ATLAS Research Theme

The ATLAS Research Theme project is an AI-assisted classification system designed to interpret free-form research descriptions and map them to LAS-approved academic fields and subfields.

This tool was developed as part of my internship with the Applied Technologies for Learning in the Arts & Sciences (ATLAS) at the University of Illinois Urbana-Champaign.

---

Overview

The College of Liberal Arts & Sciences has 70+ majors, with faculty whose research often spans multiple interdisciplinary areas. But there is no unified way for LAS staff to take a raw research description and translate it into the structured taxonomy the college uses.

The Research Theme project solves that.

It takes a plain-language query from a faculty member and returns:

• The most relevant research fields
• A refined list of subfields within each field
• Short descriptions for each selection
• A validated output that matches the official LAS taxonomy

This gives LAS staff a shared language for classification and helps with faculty placement, onboarding, and cross-college comparisons.

---

Features

1. Free-form Query Intake

Users describe their research interest however they want. This becomes the starting point for the entire pipeline.

2. Field Classification

The system analyzes the text and identifies the most relevant high-level academic fields.

3. Subfield Classification

Within a chosen field, the system drills down into more granular subfields for a sharper classification.

4. Output Validation

Every result is checked against LAS’s official taxonomy to keep classifications consistent and accurate.

5. Extensible Knowledge Base

Field and subfield data live in editable JSON files so staff can update classifications as academic areas evolve.

---

Setup Instructions

1. Clone the repository

git clone https://github.com/sharmaaaryan4012/ATLAS-Research-Theme
cd ATLAS-Research-Theme

2. Install dependencies

pip install -r requirements.txt

3. Configure your environment

• Create a file named api.env in the root directory.
• Add your Gemini API key to it in the following format:

GEMINI_API_KEY=your_api_key_here

• The app will detect this environment variable and use Gemini for LLM responses.

4. Run the pipeline

python3 main.py

You can pass any free-form research description to see the full classification flow.

---

How It Works

1. User Request The system accepts a raw research description.

2. Field Classification An LLM ranks the most likely fields based on LAS mappings.

3. Subfield Classification For each field, another LLM ranks relevant subfields.

4. Validation Results are checked against the official taxonomy for accuracy.

---

Future Directions

• Expand the tool to support inter-college research classification across the entire university.
• Enhance and maintain the knowledge base so it evolves with LAS academic priorities.
• Integrate deeper workflows for onboarding, cross-department matching, and faculty profile generation.

---

Author

Aaryan Sharma Kirthi Shankar Developed during ATLAS Internship, Fall 2025 University of Illinois Urbana-Champaign