Wikipedia-Graph-Net

The WikiRacer is a game where one tries to navigate to one Wikipedia article from another in the fewest number of clicks possible. This program's goal is to accomplish this task given the start and goal article by creating a graph and searching through it optimally using an A* search heuristic.

Test Ideas

"Dog" and "Aardwolf" "Cell nucleus" and "Cell biology"

Heuristic

Because of the semantic nature of the heuristic, we have developed multiple heuristics in the form of heuristic_# where # represents an int. Currently, the default heuristic is a cosine similarity score between the document-term matrices generated from the article intros.

How It Works

Essentially, a graph of Article classes are generated dynamically from the start point and the A* algorithm employs a priority queue (min-heap implementation) in order to efficiently traverse the graph and look for the desired end article.

Set Up

1. Clone the repository onto your local machine.
2. Navigate to the folder
3. Set up the correct venv:

   python3 -m venv venv
   virtualenv venv
   pip install -r requirements.txt
   

Make sure that you are inside the venv when you are running pip install
4. Activate the venv by running source venv/bin/activate
5. Use flask run to run the flask app and navigate to the local host