A Chat Recommender System for COVID-19 Support based in Textual Sentence Embeddings

With the emergence of the COVID-19 pandemic, the demand for health services has exponentially increased, which caused the saturation of hospital beds and a high death toll. Motivated by the need to provide more agility in patients' attendance and unburden the health services, this work proposes a solution for automatic attendance via a Recommender System that uses sentence embeddings of text messages to train an LSTM classifier. This classifier can provide recommendations of a course of action for patients, instructing them to stay at home or seek medical support. Our numerical results validate the proposed solution and corroborate its reasonable accuracy rate. © 2021 ACM.