A Micromodule Approach for Building Real-Time Systems with Python-Based Models: Application to Early Risk Detection of Depression on Social Media

Rodrigo Martínez-Castaño, Juan Carlos Pichel, David E. Losada, Fabio Crestani

Abstract: In this work we introduce Catenae, a new library whose main goal is to provide an easy-to-use solution for scalable real-time deployments with Python micromodules. To demonstrate its potential, we have developed an application that processes social media data and alerts about early signs of depression. The architecture has the following modules: (1) a crawler for extracting users and content, (2) a classifier pipeline that processes new user contents, (3) an HTTP API for alert management and access to users’ submissions, and (4) a web interface.