The intersection of physics and machine learning has given rise to a paradigm that we refer to here as physics-enhanced machine learning (PEML), aiming to improve the capabilities and reduce the individual shortcomings of data- or physics-only methods. Naturally, there exist many different motivations and schemes within the broad-scoping genre of PEML, and, recently, classification of such schemes has matured beyond the description of a one-dimensional spectrum, to a two-dimensional spectrum of physics and data [1,2], or even a three-dimensional representation, with the third dimension accounting for the involved model/algorithmic complexity [3]. Classification of schemes in this context is useful for a high-level understanding of the nature of each scheme, and can aid in appropriate scheme selection. However, in application, selection of the most appropriate PEML scheme would require users to consider lower-level characteristics, such as data domain, downstream task, prior knowledge, etc. Ontologies provide a schematic approach to objectively defining an intricate system of concepts, including interdependencies and interactions between ontological objects, and can be useful for algorithm or methodology selection [4]. This article introduces a novel ontological framework designed to provide a structured and comprehensive description of PEML methodologies. The aim is to develop a tool that assists researchers and practitioners in navigating the intricate landscape of model selection, facilitating identification of the most appropriate machine learning schemes for specific datasets, available prior knowledge, and the overarching objectives. The proposed ontology leverages principles from both the machine learning and physics domains to support decision-making processes by offering insights into the compatibility of various machine learning techniques with specific physical phenomena, allowing researchers to make informed choices that align with the nature of the data and the desired outcomes.