Latent class analysis by regularized spectral clustering
Two new algorithms for estimating the parameters of a latent class model for ordered categorical data with polytomous responses are introduced based on a newly defined regularized Laplacian matrix derived from the response matrix.