We explain how semantic hyper-graphs are used to describe ontological models of morphological rules of agglutinative languages, with the Kazakh language as a case study. The vertices of these graphs represent morphological features and the edges represent relationships between these features. Such modeling allows nearly one to one translation of the morphology of the language into object-oriented model of data. In addition, with such a model we can easily generate new word forms. The constructed model and the dictionary generated with it are freely available for research purposes.