Search

At present, the tools of information and communication interaction of the municipal and federal authorities with the citizens and organizations are actively developing. The increasing volume of electronic information leads to the need to classify multiple incoming messages. However, the specific features of such documents (small volume, lack of structuring, presence of grammatical and syntactic errors, thesaurus non-stationarity, etc.) make their statistical analysis more difficult. Also, a significant difference in the conditions of their processing does not allow using a universal method of the text document classification. This raises the urgent task of developing a multimethod approach to the rubrication of unstructured electronic documents based on the application of probabilistic and intelligent methods of analyzing text data. The computational experiments carried out in the context of interrelated and non-interrelated rubrics showed the prospects of their practical application.