Matrix text models. Text corpora models

The models of text corpora, formed on the basis of the matrix model of texts in natural languages, are presented. As methods to form models of collections we consider the techniques of computational identification of the thematic structure of the collections. We suggest to use the models for searching for thematically similar text collections and thematic categorization of texts based on text models and text collections. The differences of the proposed models of text collections from the common approaches to their analysis and modeling are analyzed.