Answers are grounded in this paper and cite their sources.
The subject of the research is the problem of objective monetary valuation of archival documents, which, despite its legal and financial significance, remains poorly formalized and largely subjective. The aim of this study is to identify the factors that determine the market value of archival documents using machine learning (ML) methods. Special attention is given to analyzing the influence of various characteristics of lots, including pre-sale expert assessments, temporal and institutional factors, as well as the content of descriptions. The work thoroughly examines the stages of dataset formation and preprocessing based on information about auctions conducted by Sotheby's. The results of applying various ML algorithms for regression and classification tasks are analyzed, and the interpretation of model predictions is performed using the SHAP method. The research methodology is based on the application of ML algorithms and natural language processing methods to predict price, valuation error, and the probability of sale. The outcome of the work is the creation of an auction sales dataset comprising 63,167 records and training algorithms to predict the market value of archival documents. The area of application of the results is the improvement of the methodology for monetary valuation of archival documents in the context of legally significant procedures. The scientific novelty lies in the quantitative assessment of the contribution of various factors to the pricing on the market for archival documents alongside traditional pre-sale expert assessments. The study demonstrated that pre-sale assessment is the main predictor, and ML algorithms can only serve as a support tool rather than a replacement for the expert. At the same time, key limitations were identified: the models are subject to regression to the mean effects (overestimating cheap lots and underestimating expensive lots), and demonstrate low accuracy in predicting the error of expert evaluation and identifying unsold lots.
No takes yet. Share an insight, caveat, or question.
Alexander Vyacheslavovich Pavlov (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: