A.Yu. Novikov, M.G. Stolyarov

Automatic text processing as it is represents a four-level model containing a graphematic, morphologic, syntactic and semantic analysis. The difficulty of solving the problem of natural language texts processing is preconditioned by non-use of formal grammar and other "strict" syntactic theories for its realization. That is why the majority of the existing solutions realize completely only the first two levels of the above-mentioned foul-level model and the remaining levels are not taken into account. These systems are intended for achieving specific aims. It is necessary to use all four levels of the model to solve completely the problem of automatic text processing. The article contains the algorithm and models of procedures of initial semantic interpretation of clauses that is a basis for the initial semantic analysis (the fourth level).