What is mmx?

MMX - a simple AI-powered meme analysis tool.

example app

backend api

• Scrapes memes periodically and builds time-resolved meme database:

  • reddit.com module

• Offers API endpoints with meme analytics:

  • lists most popular memes with highest level of engagement
  • clusters similar memes using ML techniques (cosine similarity + feature extraction)
  • finds trends

Requirements

Before building mmx, ensure you have:

• docker
• git-lfs
• a working Atlas MongoDB database (you can create a free db here); important: Atlas MongoDB is required to run the clustering algorithm)

How to start your own mmx?

To start mmx... clone the repo first.

Then, run the build script in development mode:

./build.sh dev

which creates mongodb_url - a file where the mongodb database url is stored. You need to add your database here.

Start the containers:

./run.sh dev all

Three containers are created:

• api (serves the webAPI to http://localhost:20410/api/v1/ via werkzeug)
• scrape (handles the scraping)
• feat_extract (handles the ML part)

Production

Starting mmx in production mode

./build.sh prod

gives an additional nginx docker serving as a reverse-proxy. Gunicorn is the WSGI server of the api docker.