Synchronous semantic projection of structured text onto a slide model

Abstract

This paper focuses on the applied scientific problem of mitigating potential cognitive load during presentation design by transforming linearly structured text into a slide-based model. It is justified that the prevailing WYSIWYG paradigm forces authors to simultaneously manage content formulation and spatial layout, whereas declarative generators typically operate in a “blind” compilation mode complicating the subsequent editing of the output. A method of synchronous semantic projection is proposed. This method treats the input text as an informational model and continuously maps it onto the spatial structure of a slide.  A system of slide roles has been constructed based on atomic syntactic elements and their permissible combinations. The semantic projection function “Π” is formalized to map text blocks onto the PresentationML object model, along with the function “Φ” for linking the editor cursor position to the corresponding preview region. A two-tier architecture was implemented, comprising a client layer responsible for text entry and synchronous rendering, and a server layer dedicated to lexical analysis, layout typing, and native PPTX document generation. A comparative analysis against traditional WYSIWYG editors and declarative generators demonstrates the superiority of the developed system in terms of entry threshold, potential cognitive load reduction, and post-editing flexibility of the object-level PPTX. The boundaries of the proposed method are outlined, confining its application primarily to academic and corporate presentations with a stable logical structure; complex infographics and advanced animations remain outside the current model scope.  Looking ahead, this informational model can serve as an intermediate semantic layer for autonomous report generation driven by artificial intelligence algorithms.