A method to generate random elements in GMS

In some cases, the objects of the domain space cannot be represented with appropriate feature vectors. In this case, a distance function is used to measure the similarity between the objects. In the literature there are many proposals on search structures for elements in general metric space (GMS). An important problem area in development of new storage structures in GMS is the generation of corresponding set of sample elements from the object space. The paper gives a comparison of the different representation methods and the distance matrix approach is analyzed in details. For this representation, a sample point generation method is proposed. The efficiency of the method is demonstrated with test cases.