Unbiased Bayesian

This project is a small proof of concept to help people kickstart and use a baive bayesian algorithms and natural language processing to identify texts, sentiments, and document types based on its content.

To run it as a web service, you should pre-install the following software.

• Python 3.5+
• Foreman (the ruby gem) or Honcho (the python package) if you are running in development environment
• Virtualenv

After installing these dependencies, create a virtual environment on the project root folder.

$ virtualenv -p python3 venv

Activate the virtualenv and install all dependencies listed in requirements.txt

$ source venv/bin/activate
$ pip install -r requirements.txt

Environment Variables

Declare the following environment variables before running any task on analyzer service:

• ANALYZER_DATA: Absolute path for data file (created if not exists on service bootstrap).
• API_KEY: Salted API Key that will authorize external services call the analyzer and classifier endpoints.

Create a file named .env in the root project folder exporting the environment variables to avoid environment clashing running foreman or honcho, and load it with "source" everytime you need to run it locally or on command line.

Example .env file:

ANALYZER_DATA=data
API_KEY=[API key to be used by requests]

Running the Service

To run the service in development mode, just spawn a web process with foreman or honcho.

Starting with Foreman

$ foreman start

Starting with Honcho

$ honcho start

You can install honcho on the given virtual environment to avoid additional dependencies (like ruby, ruby-gem and foreman).