Young Han, C. Park, and Key-Sun Choi
A grammar network that resembles the RTN, but takes the topology of Markov Model is introduced. This new grammar model captures the linguistic patterns from the stochastic process of the patterns. The linguistic patterns may be derived with minimal human involvement. When grammar is viewed in Markov process, we are given a rich set of tools that can be applied to the analysis and generation of languages. In this paper an informal introduction to the model is made and evaluation problem is discussed in detail. An experiment shows that the Markov nets with stack can properly model natural languages.