ML2P Filter

Greece Project type: large Liquid Media SA


Project to develop an advanced user generated content filtering system, which aims to capture all forms of profanity/harassment (e.g. irony, sarcasm, threat, insult) through machine learning. Such content is occasionally offensive, and currently moderated manually and/or using technical solutions that are based on simple term lists, missing-out contextual harassment.

The solution

Machine learning to filter profanity (ML2P filter) in user generated comments on news stories and articles.