<p>Automatic text summarization has become essential for managing the explosive growth of textual information. While several surveys exist, most concentrate on studies prior to 2019, leaving a gap in systematically reviewing the rapid advances in deep learning–driven summarization. This paper addresses this gap by providing a comprehensive review of challenges, methodologies, and emerging directions in automatic text summarization. We categorize summarization techniques into six dimensions: Input Document Number, Summary Language, Output Requirements, Summary Content, Domain Applicability, and Semantic Relationship, with particular emphasis on extractive and abstractive methods. Specifically, we synthesize nine representative extractive approaches and three abstractive approaches developed over the past seventy years, and present performance comparisons across multiple datasets and evaluation metrics. Finally, we identify seven promising future research avenues. This review offers systematic references and practical guidance for advancing research and innovation in automatic text summarization.</p>