https://github.com/stefanpietrusky/iec
Repository for the article in the online magazine Level Up Coding
https://github.com/stefanpietrusky/iec
beautifulsoup chatbot flask-application llama ollama python tiktoken
Last synced: about 1 month ago
JSON representation
Repository for the article in the online magazine Level Up Coding
- Host: GitHub
- URL: https://github.com/stefanpietrusky/iec
- Owner: stefanpietrusky
- License: apache-2.0
- Created: 2025-03-03T22:16:10.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-09-30T14:39:42.000Z (9 months ago)
- Last Synced: 2025-09-30T16:28:57.425Z (9 months ago)
- Topics: beautifulsoup, chatbot, flask-application, llama, ollama, python, tiktoken
- Language: Python
- Homepage: https://medium.com/gitconnected/iec-v1-5-individual-educational-chatbot-f859456ec760
- Size: 34.6 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
- Citation: CITATION.cff
Awesome Lists containing this project
README
# IEC V2 APP
## Individual Educational Chatbot
This repository, developed by Stefan Pietrusky, is based on the article published at [[1]](https://medium.com/gitconnected/iec-v1-5-individual-educational-chatbot-f859456ec760). In this article, I describe the functionality of an enhanced version (V1.5) of the IEC application. The first version has already been tested and provided important results for improvement.
The function of controlling the response level of the model used by clicking a button was very well received by the test subjects [[2]](https://arxiv.org/abs/2412.16165). In the extended application, the Gradio framework is no longer used, but the interface is implemented with HTML, CSS and JS. The level prompts have been optimized. The extraction of content has also been optimized. This is an open source project for educational and research purposes.
Unlike IEC V1.5, which only extracted text from URLs and PDFs and answered questions to the LLM via a simple CLI call, IEC V2 forms a fully-fledged RAG pipeline: Extracted content is divided into token chunks, semantically vectorized using Ollama embeddings, and stored in a persistent FAISS index to find only truly relevant text passages per query. The frontend now offers a selection of models and skill levels, checkbox lists for multiple sources, and dynamic content and chat overviews, while on the server side, edge-tts generates high-quality MP3s and integrates them directly into the chat with play/pause buttons. The answers are created promptly for each source, formatted with references, and merged into a consolidated overall answer, and every conversation, including timestamps, files used, and audio files, is logged completely in JSON logs.
> **⚠️ Work in Progress:** This prototyp is currently under active development. While I make it available for research purposes, please be aware that there will be some changes to the functional structure. I recognize that some current technical design decisions may not be optimal and are subject to revision. Researchers using this prototyp should expect potential updates and changes. I recommend checking back regularly for updates and versioning information.
## IEC Structure
The structure of the current [V2] IEC app is shown below.
## IEC working principle
Below is a short GIF showing the structure and function of the app.
## IEC availability
The code to run the app is already in the repository.To make the app available via the Internet and not be restricted to the local network, a tunnel tool such as ngrok must be used! Your own computer becomes the server.
Alternatively, modern Platform-as-a-Service solutions such as Railway can be used to publish the app quickly and easily.
## Installing and running the application
1. Clone this repository on your local computer:
```bash
git clone https://github.com/stefanpietrusky/iecv.git
```
2. Install the required dependencies:
```bash
pip install -r requirements.txt
```
3. Install [Ollama](https://ollama.com/) and load the default model [Llama3.2](https://ollama.com/library/llama3.2) (3B). Alternatively, another model can be used.
```bash
ollama pull llama3.2
```
4. Install the embedding model:
```bash
ollama pull nomic-embed-text
```
5. Install Python [3.10.11](https://www.python.org/downloads/release/python-31011/).
6. Adjust the Edge TTS [voice](https://gist.github.com/BettyJJ/17cbaa1de96235a7f5773b8690a20462) as needed.
7. Start the IEC app:
```bash
python app.py
```
## References
[1] Pietrusky, S. (2025). Individual learning support made easy. IEC V1.5: Individual Educational Chatbot. A tool for individual exchange with PDF files and websites. Level Up Coding.
[2] Pietrusky, S. (2024). Promoting AI Literacy in Higher Education: Evaluating the IEC-V1 Chatbot for Personalized Learning and Educational Equity. ARXIV CS.CY