Document Classification Pattern Recognition via Information Fusion: A Systematic Review of Multimodal and Multiview Representation Approaches

In three linesSystematic review of 139 studies on information fusion for document classification. Meta-analysis shows multimodal fusion improves accuracy by +5.28 percentage points (p=0.0016) and multiview fusion by +4.67% accuracy. Critical finding: only 11.8% of multimodal and 23.3% of multiview studies use statistical validation, undermining reproducibility.

Summary generated by Claude — human-verified