The field of Recommender Systems is one of the subareas of Computer Science that studies user-centered systems. My Ph.D. project had the goal of the definition of a Recommender System able to detect and consider the users’ feelings (empathic system) for supporting decisions. It involves the use of artificial intelligence algorithms for analyzing user behaviors and understand user feelings. Strategies based on natural language processing (NLP), deep machine learning, and intelligent information seeking have been used for processes of opinion mining and recommendation. Experiences about user modeling for operations of profiling and personalization have been the basis of my project. Over the past year, I have been working mainly on NLP techniques for extracting user's features from the text, including intent discovery and aspect-based sentiment analysis. In particular, I am using deep learning techniques to improve the performance of text classification strategies, mainly on Italian language and hate speech topics.