This project was conducted for a mobile-first marketing company that analyses exclusive online and offline data sources at scale to provide brands with powerful insights on their audiences for a more accurate media activation.
Our project team was composed of Rémi FERREIRA, Maxime POULAIN, Olivier SCHULTZ and myself.
The goal of this project was to build a multi-label classifier capable to assign between one and five thematic tags — from a bank of 1.300 possible tags — to a web page, based on its content.
Input would be the URL of the web page to classify. In production, the classifier would have to process 50.000 URLs every 45 seconds.