Near-Infrared Spectroscopy and Machine Learning for Accurate Dating of Historical Books

• DOI: 10.1021/jacs.3c02835
• Publication Date: 2023-05-22T18:17:41Z
• Authors: Floriana Coppola; Luca Frigau; Jernej Markelj; Jasna Malešič; Claudio Conversano; Matija Strlič
• Keywords: Overtone

Non-destructive, fast, and accurate methods of dating are highly desirable for many heritage objects. Here, we present critically evaluate the use near-infrared (NIR) spectroscopic data combined with three supervised machine learning to predict publication year paper books dated between 1851 2000. These provide different accuracies; however, demonstrate that underlying processes refer common spectral features. Regardless method used, most informative wavelength ranges can be associated C-H O-H stretching first overtone, typical cellulose structure, N-H overtone from amide/protein structures. We find expected influence degradation on accuracy prediction is not meaningful. The variance-bias decomposition reducible error reveals some differences among methods. Our results show two out allow predictions dates in period 1851-2000 NIR an unprecedented up 2 years, better than any other non-destructive applied a real collection.