A set of cognitive categories reflecting the process of scientific cognition is required to assess the semantic consistency of a text description. Categorical characteristics of a scientific text are implemented through a system of paradigm dispositions as a model of problem statement and decision support. In existing decision support systems, construction of conceptual frameworks of scientific texts is based on the methodology of formal systems of descriptive logic. Since it is a subset of predicate logic, its expression is limited. To level out the existing limitations, it is proposed to use higher-order logics and type theory to construct a conceptual framework of scientific texts. An algorithm for constructing a conceptual framework of scientific texts is proposed to assess the semantic consistency of a text description with knowledge of the subject area. In scientific texts of different functional styles and pragmatic orientations, the conceptual framework has different supporting categories. The substantive categories of information content and the structural categories of discreteness and coherence of the conceptual framework of scientific texts are identified.