Multidimensional Feature-Based Textbook Difficulty Assessment Index

Abstract

This study addresses limitations of existing textbook difficulty assessment methods, often relying on single-dimensional metrics. Grounded in cognitive load theory, a multi-dimensional feature index is proposed, integrating five key dimensions: linguistic complexity, formula density, diagram complexity, knowledge abstraction, and structural disorganization. The index is constructed using techniques from natural language processing, image analysis, and knowledge graphs. Linguistic features are derived through tokenization and syntactic analysis; LaTeX formulas are detected via regular expressions; diagram complexity combines structured data and image texture features; knowledge abstraction uses dynamic terminology matching; and structural disorganization is assessed through chapter detection and coherence analysis.