High-performance functional materials are the cornerstones of the continuous advance of modern science and technology, but the development of new materials is still challenging. Here, we propose a robust design strategy for novel crystalline solids based on group-theory classification and high-throughput computation, as demonstrated by the successful identification of new optoelectronic semiconductors. First, by means of theoretical group analysis and composition engineering, we obtained 78 prototypical crystal structures and built a computational materials database containing 21,060 ternary chalcogenide compounds. Our high-throughput screening of the coordination characteristics, phase stability, and electronic structures provided 97 candidate semiconductors, including 93 completely new compounds. Among them, 22 crystals with excellent dynamical and thermal stability are predicted to show high photovoltaic conversion efficiency (>30%), comparable to the currently most efficient single-junction GaAs solar cell, owing to their optimal electronic properties and outstanding optical absorption. This discovery of new chalcogenide crystals offers excellent candidates for optoelectronic applications and suggests that our design strategy is a promising way to search for unknown high-performance functional materials.