In this talk we present SpEQ, a quick and correct strategy for detecting semantics in sparse codes and enabling automatic translation to high-performance library calls or domain-specific languages (DSLs). Sparse compilers can automatically generate fast implementations for given computations using requested storage formats; however, the opposite direction – retrieving high-level computation and storage details from already sparse codes – is largely unaddressed. Sparse codes are difficult to optimize automatically without full knowledge of these computation and storage details, yet, many large programs, such as legacy codes, contain handwritten sparse implementations that are tedious and error-prone to optimize by hand. And, when the next language or DSL is introduced, the process must be repeated.