The study of decision-making has gained more and more importance, from the classical conceptions of the economic man to the more recent concepts of bounded rationality and cognitive biases. Over time, the increasing complexity of decisions has driven the development of technologies such as Decision Support Systems and Predictive Models, highlighting more recently the incorporation of techniques from the field of Artificial Intelligence, and more precisely Machine Learning, to improve the accuracy and efficiency of decision-making. However, as great as the benefits provided by advances in computer support have been, humans are ultimately the ones to make decisions. And, being an essentially human task, the influence of cognitive biases on decision-making is a relevant and underexplored challenge. These biases can be due to various factors, including individual preferences, external influences, and unconscious cognitive derivations. Despite the efforts of the field of Behavioral Economics to identify and model these biases, their impact in contexts of monetary decisions is still limited. Therefore, this work proposes an architecture based on ontological foundations to identify and analyze cognitive biases in scenarios of high risk of monetary losses. Through the application of Data Science and Machine Learning techniques, we propose a methodology - implemented in a computational artifact - capable of automatically identifying patterns of cognitive biases from a history of decision records, generating knowledge about the risk preferences of decision makers and their gains and losses caused by their choices. The specific bias explored in this study is Risk Seeking in the loss domain, as defined in the Kahneman Quadruple Pattern. The evaluation of the effectiveness of this proposal will be carried out through a case study using a benchmark available in the literature, providing insights into the applicability and practical benefits of the proposed architecture.