The complexity of data and methods in industrial ecology (IE) keeps growing, and the demand for comprehensive and interdisciplinary assessments increases. To keep up with this development, the field needs a data infrastructure that allows researchers to annotate, store, retrieve, combine, and exchange data at low cost, without loss of information, and across disciplines and model frameworks. A consensus-building debate about how to describe the common object of study, socioeconomic metabolism (SEM), is necessary for the development of practical data structures and databases. We review the definitions of basic concepts to describe SEM in IE and related fields such as integrated assessment modeling. We find that many definitions are not compatible, are implicit, and are sometimes lacking. To resolve the conflicts and inconsistencies within the current definitions, we propose a hierarchical system of terms and definitions, a practical ontology, for describing objects, their properties, and events in SEM. We propose a typology of object properties and use sets to group objects into a hierarchical, mutually exclusive, and collectively exhaustive (H-MECE) classification. This grouping leads to a general definition of stocks. We show that a MECE representation of events necessarily requires two complementary concepts: processes and flows, for which we propose general definitions based on sets. Using these definitions, we show that the system structure of any interdisciplinary model of SEM can be formulated as a directed graph. We propose guidelines for semantic data annotation and database design, which can help to turn the vision of a powerful data infrastructure for SEM research into reality.