Abstract: Hydraulic fracturing is of critical importance in tackling gas control challenges in deep, low-permeability coal-measure reservoirs, and a comprehensive understanding of the propagation mechanisms, characterization approaches, and effectiveness evaluation methods for hydraulic fractures is essential for achieving permeability enhancement and gas extraction improvement through fracturing. From a "characterization-control-evaluation" perspective, this review systematically examines the current state of research on hydraulic fracturing in gas control. In addition, it establishes a fracture characterization framework based on multi-source information integration and intelligent inversion, and subsequently proposes a comprehensive full-cycle evaluation system for fracturing-enhanced permeability and gas extraction oriented toward gas control by combining multidimensional indicators covering fracture morphology, permeability enhancement, extraction performance, and engineering benefits. Findings reveal that existing fracture propagation theories lack sufficient adaptability to highly heterogeneous reservoirs. Consequently, fracturing parameters should be optimized according to geological conditions and adjusted dynamically with real-time monitoring. Given the inherent limitations of individual fracture characterization techniques, integrating multi-source, cross-scale monitoring data is necessary for developing reliable fracture models. To objectively assess the positive impact of hydraulic fracturing on gas control, a multidimensional evaluation system spanning from process to outcome and from microscopic to macroscopic scales should be established. For future intelligent transformation of hydraulic fracturing, promising directions include developing digital twins for hydraulic fracturing and gas extraction, establishing effectiveness mappings between hydraulic fracturing and gas control outcomes, and implementing adaptive closed-loop control systems. These advances will address key scientific challenges, such as transparent cognition, accurate prediction, and optimized control, and promote a paradigm shift from experience-based practices to quantitative scientific engineering.