The document describes a method for justifying recommendations through aspect-based sentiment analysis of users' reviews. It involves extracting aspects from reviews using natural language processing, ranking aspects by relevance and sentiment polarity, and generating a natural language justification using positive excerpts about high-ranking aspects. An experimental evaluation with 286 subjects compared justifications from different combinations of parameters and to a feature-based baseline. Results showed that review-based justifications scored higher than the baseline in terms of transparency, persuasion, engagement, trust and effectiveness.