Small experiments with sentiment analysis on twitter. This will listen to tweets of specified topics and run sentiment analysis on them. The data is stored in a Postgresql database. The app also hosts an HTTP server, to be run behind a reverse proxy, which shows the data with some nice SVG graphs.
Rename .env.sample to .env and config.sample.yaml to config.yaml and fill in appropriate values. Make sure the Postgres server is running. Then simply run with cargo run --release.
When trying to run the binary without cargo, it usually fails to find the libtorch libraries. Set LD_LIBRARY_PATH to the proper folder to resolve this.
Alternatively, build a docker image with docker build -t repo/tag .. Make sure the docker container for this image has access to the environment variables, the config file and the postgres server.