A framework for conceptualizing data properties as input to the construction of facts

Abstract

Data-related literature is permeated with fuzzy, multi-dimensional and ill-defined concepts like big data leading to low
theoretical or practical usability to actors trying to succeed in an increasingly complex data environment. To
successfully construct facts from data, a data language is needed which instead is grounded in well-established, precise
and unambiguous concepts and terminologies. To contribute to this objective, this conceptual study works out the
multiple data characteristics described in literature to clearly define those, to set unambiguous label names (data classes)
for them, and to classify them into data dimensions representing specific meta-perspectives relevant to actors. By
reviewing a wide array of data-related literature and the use of five classification principles, 63 data classes and 23 data
dimensions have been identified. They have, sometimes in detail, been discussed, labelled and defined using the three
dimensions of meaning from pragmatic constructivism, and they have been located inside a data value model which
illustrates the key processes between a measured phenomenon and data practices. The resulting classification
framework provides a shared understanding of universal data concepts and contributes to a more unambiguous data
language in theory and practice.