Various generalized dual-hard-sphere (DHS) models are reviewed on calculating the liquid structure factor for semiconductor elements Si and Ge. It is found that the model generalized by Canessa, Mariani and Vignolo gives the best fitting of experimental structure factorSexp(k) in the rangek> 2kF, (kF, the Fermi wave vector), and all previous models including a new generalized model by the author fail to reproduce the experimental structure factorSexp(k) of Si and Ge in the whole range ofkvector.