The Large-Scale Structure (LSS) of the Universe is a homogeneous network of galaxies separated in dense complexes, the superclusters of galaxies, and almost empty voids. The superclusters are young structures that did not have time to evolve into dynamically relaxed systems through the age of the Universe. Internally, they are very irregular, with dense cores, filaments and peripheral systems of galaxies. We propose a methodology to map the internal structure of superclusters of galaxies using pattern recognition techniques. Our approach allows to: i) identify groups and clusters in the LSS distribution of galaxies; ii) correct for the "fingers of God" projection effect, caused by the partial knowledge of the third space coordinate; iii) detect filaments of galaxies and trace their skeletons. In this paper, we present the algorithms, discuss the optimization of the free parameters and evaluate the results of its application. With this methodology, we have mapped the internal structure of 42 superclusters in the nearby universe (up to $z=0.15$).