Hi and welcome to my KG project!
This application is designed to allow users to see similar flats in Willhaben using both data from their website, linked with public transportation stops from the Wiener Linien GTFS data.
User can see small number of flats on a map and can click on the to get similar flats, predicted via clustering on KG embeddings. Users can also choose to show public transportation stops in Vienna.
The data folder should contain all data necessary to run the project as is.
This includes an exported edge-list from Neo4J, the pykeen model, the k-means clusters and the Willhaben scrape + Wiener Linien data.
For the backend, you will need to install a number of Python dependencies. I recommend doing this via Conda, using the provided environment file environment.yml.
In order for the project to work, devs will need to install Neo4J locally.
The setups requires the following credentials to work:
URI = 'bolt://localhost:7687'
AUTH = ('neo4j', 'kg2023summer')
DB = 'neo4j'
If you want to change the connection parameters, please navigate to helpers/neo4j_driver and adapt them!
For the frontend, you will need to install Vue 3 in Vite. I recommend doing this via the Node package manager npm. In theory nodejs should be included in the Conda environment, but it may be weird, so check if node is installed and then install Vue 3 in Vite.
The backend runs at localhost:8000 and the frontend at localhost:5173. You can change them easily if you want, but beware that a few changes to the CORS configs may be necessary.