Welcome! This bot comes in 3 parts, which can all be run on the same machine, or even different machines!
Head over to the binary classification repo and clone it.
Inside is one relevant file. That file compiles a machine learning model from the data you provide it, and then launches a server to allow the bot to interface with the machine learning model.
It contains no authentication, do not port forward... obviously.
At the top of the file are various helpful comments explaining the required directory structure for training. At the project root level is the folder "suicidedata", which holds the folders "test" and "train", each of which hold the folders "suicide" and "not", each of which holds text files of your training data, with each file being an indiviual datapoint.
In the end, it would be something like this:
BinaryClassifier/
├─ main.py
├─ suicidedata/
│ ├─ train/
│ │ ├─ suicide/
│ │ │ ├─ 0001.txt
│ │ ├─ not/
│ │ │ ├─ 0002.txt
│ ├─ test/
│ │ ├─ suicide/
│ │ │ ├─ 0003.txt
│ │ ├─ not/
│ │ │ ├─ 0004.txt
For some sources of data, check out the links in the comments of main.py.
Once the folders are organized and filled, run python3 -m pip install tensorflow==2.15.0 to install TensorFlow.
If you are on macOS, ensure you are running the version of python native to your archetecture, i.e. ARM-native python for Apple Silicon, and Intel-native python for Intel.
Finally, you're go for launch. run python3 main.py and go for a walk outside. A long walk. Longer. This will take hours... assuming you don't run into any errors.
Once it's running, you can take a deep breath and move on to Part 2. You can run it through pm2 if you want it daemonized.
Clone this repository, and grab the schema.sql file.
Install mysql, and then run schema.sql through mysql to set up the database.
Make sure that the database is accessible via a password, and make sure you save that password somewhere.
If you have not done so already, simply clone the repository, and open it in your IDE of choice.
Once you have done that, duplicate the .env.example file, and rename the new version to .env. It comes with prefilled example values that you will need to change yourself. Do the same with config.json.example, renaming it to config.json.
Install node and npm if you have not done so already.
Once everything is set just right, run npm i and node run index.cjs to launch.
In theory, the program will connect successfully to both the database and the machine learning model, and it will be ready to go.