Keywords: nuclear safeguards, fuzzy linguistics, missing values, trustworthiness, imputation, nuclear safety, nuclear weapons, nuclear power, nuclear energy, nuclear security, security evaluation, uncertainty management, modelling, incomplete information, trustworthy results
A hybrid model to deal with missing values in nuclear safeguards' evaluation
Nuclear safeguards' evaluation aims to verify countries not misusing nuclear programmes for nuclear weapon purposes. This is a complex process in which the International Atomic Energy Agency (IAEA) uses information from different available sources. Initially, the research was focused on developing the evaluation framework to handle the complexity of the safeguards' evaluation and the uncertainty inherent to the information involved. Consequently, different models were obtained to cope with the complexity by means of a multi-level hierarchical process, and the uncertainty managed by modelling the information linguistically. However, another challenge in nuclear safeguards is to handle the incomplete information due to the missing values that could bias the final evaluation results. There exist different strategies to treat missing values in nuclear safeguards. This paper presents a new approach for dealing with missing values that hybridises the previous ones to increase the trustworthiness of nuclear safeguards' evaluation results.