CoolFace
Apppublic

keefereuther/Schema_Study_Preview

sourceHugging Facegpl-3.0updated 4mo agoView on Hugging Face
0likes
App README

Schema Study: An AI-Enhanced Study App for Biology Students

Schema Study is a free, no-code, open-source web app that turns a spreadsheet of course terms into an AI-powered study coach. It is designed for asynchronous student learning and inquiry: students work at their own pace and get instant, personalized feedback. The app uses evidence-based teaching practices and Socratic questioning—guiding you with questions instead of giving answers—to deepen understanding, correct misconceptions, and help you connect concepts. It is based on the research described in Schema Study: A Large Language Model (LLM) Application for Asynchronous Student Learning and Inquiry (CourseSource, 2026).

In short: You pick a term, the AI tutor (Pliny) asks you to explain it and then follows up with one focused question at a time, so you practice thinking through the material instead of just reading answers.

Getting Started: Your API Key (Required for This Preview)

This evaluation version asks you to enter an OpenAI API key in the sidebar so the chat can run. You only need to do this once per session.

What is an API key?

An API key is like a secure password that lets an application (such as Schema Study) talk to an external AI service (OpenAI) on your behalf. When you type in the chat, the app sends your message to that service; the key proves you are allowed to use it. Think of it as: “This app has permission to use my OpenAI account to generate responses.”

  • —You control it: You create the key in your own OpenAI account. You can revoke it anytime.
  • —Pay only for what you use: There’s no monthly subscription for the key itself. You add a small amount of credit (e.g. $5–10) and are charged only for the AI usage (tokens). For a full class over a term, institutional use has been very low cost (e.g. on the order of hundreds of dollars total).
  • —Keep it private: Don’t share your API key or post it publicly. If someone had it, they could use your OpenAI credit.

How do I get an API key?

Step-by-step video (recommended): We have a short walkthrough that shows exactly how to get an OpenAI API key and where to enter it:

[How to get an OpenAI API key — Video walkthrough](https://www.youtube.com/watch?v=QiJ6cWm9Dx4)

After you have your key, open the sidebar in this app and paste it into the “OpenAI API Key” field. The chat will then work. The app does not save your key after you leave.

Features

  • —No-Code Setup: Upload course terms via a single CSV spreadsheet—no programming required
  • —API Key (Preview Only): Enter your OpenAI API key in the sidebar once to enable chat; see the video for how to get one
  • —Customizable Terms: Use your own CSV file of terms and definitions
  • —Evidence-Based Pedagogy: Socratic questioning and formative feedback (withhold answers, one follow-up question per turn, focus on reasoning and connections)
  • —Prompt Templates: Five conversation starters (e.g. Misconception Check, Two Truths & a Lie, Connect Terms, Schema Map, Create Study Plan)
  • —AI Feedback: Instant formative feedback via GPT-5.2 (default), GPT-5.1, or GPT-4.1
  • —Web Search (Optional): Configurable in config.py for current information and citations
  • —Streaming & UI: Live streaming responses and a clear, accessible interface
  • —Open-Source & Free: GNU GPL-3; no paywalls

Evaluation Versions

This is the evaluation/preview version of Schema Study—for trying the app before you duplicate and customize it.

On these previews, you must enter your own OpenAI API key in the sidebar to use the chatbot. For a step-by-step guide, see [How to get an OpenAI API key](https://www.youtube.com/watch?v=QiJ6cWm9Dx4).

Production use: To run Schema Study for your course (with your own terms, optional password, and no student API keys), use the production Space and the user manual: Schema_Study and S2_user_manual.pdf.

How to Use (Students)

  1. 1.Open the app and enter your OpenAI API key in the sidebar (see video if you need one).
  2. 2.Choose a term from the dropdown.
  3. 3.Study: Answer the prompt or use a template. Ask questions, explain ideas, and follow Pliny’s follow-up questions to go deeper.

How to Use (Instructors)

Prerequisites

  • —A web browser (Chrome, Firefox, or Safari)
  • —An email address
  • —A credit card to add about $5–10 of OpenAI credit (you pay only for actual usage)
  • —Your course syllabus or list of learning objectives
  • —About 1–2 hours to create and customize your own copy of Schema Study

Setup Instructions

This Space is a preview. Here, each user enters their own OpenAI API key. For a short visual guide on getting an API key, share this with students: [How to get an OpenAI API key](https://www.youtube.com/watch?v=QiJ6cWm9Dx4).

To build your own version for your course (with your terms, optional password, and no student API keys), use the [Schema Study User Manual](S2_user_manual.pdf) and the production Space: Schema_Study.

The user manual covers:

  • —Setting up Hugging Face and OpenAI accounts (and getting an API key)
  • —Creating and customizing your terms CSV
  • —Configuring your Hugging Face Space
  • —Customizing app settings in config.py
  • —Sharing and embedding Schema Study in your LMS
  • —Model selection and pricing

Configuration

AI Model Settings (config.py)

  • —Default Model: GPT-5.2 with reasoning="none" for faster responses
  • —Alternative Models: GPT-5.1 (reasoning model) or GPT-4.1 (non-reasoning model with temperature control)
  • —Web Search: Configurable via enable_web_search (default: True)
  • —Reasoning Effort: Configurable for GPT-5.2 and GPT-5.1 (options: "none", "minimal", "low", "medium")
  • —Temperature: Configurable for GPT-4.1 (0.0-2.0)

Pacing and Question Control

Schema Study enforces structured pacing to prevent cognitive overload. The system prompt enforces one focused question plus one short, realistic application example per turn to reduce confusion from multiple simultaneous follow-ups.

Universal Design for Learning (UDL) Adjustments:

  • —For additional scaffolding: Modify term_prompt to allow brief clarifying questions
  • —For advanced students: Add instructions to keep responses focused and move efficiently toward connections

The five prompt templates provide structured entry points for students who need additional support initiating conversations.

Other Settings

  • —All settings are in config.py (title, instructions, prompt templates, resources, AI model parameters, etc.)
  • —Theming is managed via .streamlit/config.toml and custom CSS in app.py
  • —Dependencies are listed in requirements.txt

Technical Details

API & Models

  • —API Framework: OpenAI Responses API (streaming-enabled)
  • —Supported Models: GPT-5.2 (default), GPT-5.1, GPT-4.1
  • —Streaming: Real-time token-by-token response streaming
  • —Inactivity Guard: Streaming stops after 60s of no server deltas

File Structure

  • —app.py — Main Streamlit app with Responses API integration
  • —config.py — All app settings and customization (model selection, web search, prompt templates, system prompt)
  • —requirements.txt — Python dependencies
  • —terms.csv — Your course terms and definitions (CSV format: term, context)
  • —example_syllabus.pdf — Example resource file (replace with your own syllabus)
  • —LICENSE — GNU GPL-3 License file
  • —S2_user_manual.pdf — Comprehensive user manual with detailed setup and configuration instructions

License

This project is licensed under the GNU GPL-3 License. See the LICENSE file for details.

Best Practices for Integration

Based on classroom testing and iterative refinement, here are recommended best practices:

  • —Clear Structured Messaging: Provide students with clear instructions on how to use Schema Study and its role in your course
  • —AI Literacy Training: Include brief training on AI literacy and appropriate use of AI tools in educational settings
  • —Curricular Alignment: Ensure your terms and context align with your learning objectives - update content regularly as you progress through the course
  • —Scaffolded Active Learning: Embed Schema Study within structured assignments rather than as an optional tool
  • —Formative vs. Summative: Use Schema Study for formative practice and feedback; evaluate independent performance in secure assessments outside the app

Research & Citation

Schema Study and its manuscript were authored, edited, and tested by Keefe Reuther, Liam O. Mueller, Grace Constantian, and Albert Nguyen.

The app was developed to address a central challenge in undergraduate biology: giving immediate, personalized formative feedback in large, diverse classes. It uses evidence-based teaching practices and Socratic questioning to deepen understanding, correct misconceptions, and help students connect course concepts.

Evidence from the classroom: In Winter 2025, Schema Study was used in an introductory biology course (N = 225). Pre- and post-surveys showed strong satisfaction: 72% of students said they would reuse Schema Study in future biology courses. More frequent use was associated with a much higher likelihood of recommending it. The tool also increased students’ AI self-efficacy and their sense that AI is relevant to their education and careers.

If you use Schema Study in research or teaching, please cite the paper:

*Reuther, K., Mueller, L. O., Constantian, G., & Nguyen, A. (2026). Schema Study: A Large Language Model (LLM) Application for Asynchronous Student Learning and Inquiry. CourseSource 13.*

The full article is available at CourseSource (see Latest) and in the CourseSource collection on QUBES. (Note: The DOI cited in the manuscript may not yet be active.)

Production version: Schema_Study.

Acknowledgments

This work was supported by University of California, San Diego intramural grants TG114333 and RG113974.

Support

For questions about creating your own version of this application for use in your classroom, please email kdreuther@ucsd.edu.