Skip to content

Latest commit

 

History

History
64 lines (53 loc) · 2.17 KB

README.md

File metadata and controls

64 lines (53 loc) · 2.17 KB
Raw

Scholar-s-Ally: An extractive/Generative QA solution designed to navigate and distill complex educational materials.

Extractive Question Answering (QA) is a type of question-answering system where the answer to a question is extracted directly from a text or a set of texts. In extractive QA, the answer is a segment of text (such as a phrase or sentence) taken verbatim from a document or documents, rather than being generated or summarized. This method relies on identifying and extracting the most relevant parts of the text that directly answer the posed question.

In generative QA, the answer to a question is not extracted directly from a pre-existing text but is instead generated by the system based on its understanding of the context and the question.

Scholar's Ally. How it works?

  1. Integrate educational content: chapters of a specific subject at school (pdfs).
  2. Type a question about it.
  3. Get the most relevant information from the integrated content and where to find it.

Example Image

Architecture

Architecture Image source: https://docs.cloud.deepset.ai/docs

Installation steps

1. Clone the repository:

git clone https://github.com/alaeddinehamroun/Scholar-s-Ally.git
cd Scholar-s-Ally

2. Create a virtual environment and activate it:

# Make sure you have python3 and pip installed
python3 -m venv .venv
source .venv/bin/activate

# To  deactivate the environment run:
# deactivate

3. Install the required dependencies:

pip install --upgrade pip
pip install -r requirements.txt

4. Launch Elasticsearch server:

You can visit the docs of Elasticsearch for more details.

# Make sure that Docker is available in your environment.
python3 src/launch_es.py

or

docker pull docker.elastic.co/elasticsearch/elasticsearch:7.9.2
docker run -d -p 9200:9200 -e "discovery.type=single-node" elasticsearch:7.9.2

5. Start the API:

cd src/api
uvicorn main:app --reload
  1. Index: to index your files availables at data/ folder
  2. Query: enter your question and you'll get the answer

6. Start the UI:

cd src/
python3 -m streamlit run ui/home.py