Comparación de modelos novedosos de proximidad en quimioinformática

This work comprises the computational implementation in the Java environment of 21 proximity models to be used in simulated experiments of similarity searching, nine out of which are novel in Chemoinformatics since they come from the psychology field, and other 12 are measures already established in the specialized literature. Afterwards, the similarity measures were compared and assessed at the "early retrieval" using nine data sets from medicinal chemistry, represented by machine learning-selected real descriptors, and one efficient matching algorithm. Results show that in average trends the new models perform superiorly with respect to the reference ones, and more than half of them are among the top-10 models.