Vectors are massively used in many domains. Several techniques have been proposed for comparing two vectors, but they only perform the comparison according to the exact values of the vector components. Additionally, existing techniques used for comparing two vectors having different dimensions are limited by many factors. Furthermore, the problem of comparing two finite sets of vectors has not yet been specifically addressed. This paper attempts to overcome all these limitations by proposing a new technique based on hidden Markov models which enhances existing techniques by giving them the ability to compare two finite sets of vectors, each containing vectors having different dimensions, while precising the set of targeted properties on which the comparison should be performed. Classification experiments conducted on three online available custom datasets demonstrated that when the suitable set of targeted properties is selected, the proposed approach outperforms existing techniques with accuracy gains reaching +82.3%. INDEX TERMS Vectors, vector comparison, distance between vectors, hidden Markov models.