Skip to content

Cluster and visualize relationships between anime on MyAnimeList

Notifications You must be signed in to change notification settings

platers/MAL-Map

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

75 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MAL Map

Cluster and visualize relationships between anime on MyAnimeList.

Description

MAL Map is a web application that visualizes anime relationships on MyAnimeList and AniList. Edges are extracted from the recomendations of users.

Clustering

The graph is clustered with multi-level modularity clustering. Some clusters are merged to produce a simpler visualization.

Layout

The map layout is generated by a particle force simulation. At first all of the nodes are represented by a single particle. The simulation is then run for a number of iterations. Each iteration all particles are split into new particles, one for each sub-cluster. Appropriate forces are applied between particles to attract related clusters together. This process results in a visually pleasing layout. The layout process can be seen live here.

Contributing

Pull requests are welcome!

Building

MAL Map is two projects: a backend node app for collecting data and clustering, and a svelte frontend client. The frontend depends on some files in the backend and its outputs.

Building the backend

  1. Open the /data-collection directory and run npm ci to install all dependencies.
  2. Python3 is used for clustering. Install networkx and cdlib with pip install -r requirements.txt.
  3. Run npm run reset to build and run the whole data pipeline. This will pull all data from MAL/Anilist and create various txt and json files for the frontend. Use npm run layout to skip the data collection step.

Building the frontend

  1. Open the /svelte directory and run npm ci to install all dependencies.
  2. Run npm run dev to start a localhost server with live reloading. http://localhost:8080/#animate=true will show a live view of the layout.

Acknowledgements

Clustering

Layout

Frontend

Data