Abstract
Document paper identification is important in forensic document examination for assessing the source, substitution, and authenticity of questioned documents. This study evaluated a dual-column fast gas chromatography (fast GC)–based electronic nose (E-nose) combined with chemometric analysis for document paper discrimination. Headspace volatile fingerprints were obtained from 50 document paper observations, and peak-area datasets from two chromatographic columns were analyzed separately and in combination. Tentative library-based annotations were used only for qualitative interpretation. Principal component analysis, radar plots, and hierarchical cluster analysis were applied to examine clustering patterns and discriminative retention-time variables. The combined dataset showed clearer clustering than the individual column datasets, indicating complementary information from the two columns. In internal nested cross-validation, k-nearest neighbors, support vector machine, and principal component analysis–linear discriminant analysis achieved accuracy, balanced accuracy, and macro F1-score values of 1.00 for the combined dataset. However, no independent external test set was used. These results indicate that dual-column fast GC E-nose fingerprinting may provide a rapid, solvent-free, complementary screening approach for forensic document paper discrimination, although validation with larger and independent datasets is required.