this is a WIP side-project for learning recommender systems.

this tool pulls data from kitsu, then generates a graph and a few neural network models from the data in order to find the best anime titles for you.

research from the netflix prize made it very apparent that one method isn't adequate, so we do a few different methods at once. some is to prevent cold-start issues, but turn themselves off after enough data is gathered.

you need

• Go (ideally 1.9+)
• node.js (ideally 8.4+)
• docker-compose
• go dep

# get dependencies
dep ensure

# build the binaries
./tool install

# start dev services
docker-compose up -d

# set up environment vars
eval $(./tool env)

# currently, running the services is volatile,
# so best bet is to just run based on cmd folder names.
# as the project matures, this documentation will mature with it.

most of the work was streamed on twitch, http://twitch.tv/jumpystick

also consider joining my community's discord, https://discord.gg/Z5C8JZu