The functioning of almost any organization to one degree or another depends on how reliably its information resources are protected from various information security threats, one of which is spam. At the same time there have been many attempts to solve the problem of its detection once and for all. Research is ongoing in this subject area constantly. Based on its results, various approaches are proposed and implemented in practice. The authors previously proposed a model of e-mails that takes into account the content of e-mails, which often changes depending on the tasks performed by users and their changing information needs. This article discusses the issue of preprocessing e-mail texts in the problem of spam detection using a model of e-mails obtained on the basis of a genetic approach to the formation of mathematical models of texts, which has proven itself for solving various problems.