At Emakia, our mission is to safeguard users and recipients from undesirable content on social media platforms. We are in the process of crafting a system designed to filter out harassment while offering a robust support framework for those who experience harassment. This system aims to extend mental health support to the broader online community as well.

Utilizing Machine Learning classifier models, the Emakia system processes incoming social media data to address the challenge of online harassment effectively. We developed a prototype that utilizes REST API protocols for data transfer. Our system has been trained using ML classifiers, including CoreML and AutoML, on English-language data to screen for and eliminate unwelcome content. Our prototype iPhone app, Enaëlle, employs these trained models directly on the user’s device. Additionally, an evolving lexicon acts as an adaptive filter, allowing for the retraining of the model with new and unidentified terms. To monitor our progress and refine our approach, we employ a comprehensive set of metrics.