The work develops the concept of flexible word linking of recognized text administrative documents. The issues of analyzing text documents in systems for entering and recognizing administrative documents are considered. The analysis is carried out using of two tasks: determining the type of document to be recognized and structuring it. The work discusses distance functions that should provide decomposition and solution of these problems. The implementation of document input and recognition systems may vary. The chapter proposes models of recognition results and distance functions applicable to administrative documents in which the vocabulary of static words is limited and may be highly noisy when digitized (scanned or photographed with mobile devices). Several models of recognition results have been proposed: a sequence of characters and a sequence of character location places. A character or character location may have one or more ratings. Methods of linking are based on algorithms for matching words and descriptions. The proposed concept can be applied in modern CAD systems to analyze the content of recognized textual files.