The Data, Information, Knowledge, and Wisdom (DIKW) pyramid depicts the general principle of a decision-making process. Various levels of abstraction are required to formalize the available data/information/knowledge with respect to the decision being supported. The Evaluation Model Development and Assessment Process (EMDAP) was developed by the United States Nuclear Regulatory Commission (US NRC) to guide the development and assessment of Evaluation Models (EMs) for transient and accident analysis of nuclear power plants under design basis accidents. It concerns the decision of adequacy of an EM with respect to the purpose of analysis (NPP application). However, lack of data, scaling issues, modeling limitations, computation, and experimental overhead are some of the key factors that challenge the EM adequacy decision in EMDAP. Phenomena Identification and Ranking Table (PIRT) which is used for complexity resolution can also introduce noise due to bias and variance in expert opinion and judgment. The uncertainty in the decision due to these issues and challenges can be minimized by enhancing the EMDAP from the DIKW perspective. In this work, we formulate a systematic scheme for knowledge representation, and classification and characterization of evidence to enhance clarity and transparency in the EMDAP implementation. The proposed scheme is governed by the concept of the value of information, where we contextualize, classify, characterize, and evaluate the evidence (relevant data/knowledge/information related to EM development, verification, validation, and uncertainty quantification) with respect to the relevant decision attributes or elements of EMDAP. We make use of the Predictive Capability Maturity Model (PCMM), Argumentation technique, and process quality assurance for knowledge representation to support EMDAP implementation in advanced reactor licensing applications. The main objective of the proposed scheme is to support decision-making by making it more transparent, traceable, and accountable. The illustration of the proposed scheme is provided using a system simulation code (SAS4A/SASSYS-1) as the primary EM and a generic sodium fast reactor under loss of flow accident as the target reactor application.EMDAP involves different elements that require expert input and insights. However, the process of expert elicitation may require significant resources and time. To address these issues, we identify generative artificial intelligence models as potential tool which (after systematic testing and capabilities analysis) can be used for domain specific expert elicitation and information extraction for efficient implementation of EMDAP.