7.1.4Subject classification and granularity level

The next step towards quantification and measurement of interdisciplinarity – at least for the diversity concept – is the choice of a pre-defined classification scheme and the level of granularity. In other words, one has to decide what level knowledge integration needs to be studied and how research can be assigned to disciplines defined at the selected level of granularity. In practical terms, this means that one has the choice between broader field, topic interdisciplinarity or some level in between. The choice of a higher level of granularity would allow the studying interdisciplinarity both at the global level (i.e., across all subjects or topics) as well as locally, this is, within a given field or discipline. A low granularity would only allow global-level studies of knowledge integration across major fields or large research areas. However, cognitively overlapping fields at low granularity level and inconclusive multiple-assignments at higher granularity resolution, substantiate the cognitive dimension.

Therefore, the decision on granularity cannot merely be based on theoretical or conceptual considerations as granularity is in general limited by its own cognitive characteristics of ambiguity that do not allow searching for neither a too coarse nor a too fine-grained solution without losing the underlying structure, or using Michelangelo Antonioni’s film “Blow-Up” (1966) as an allegoric picture of uncertainty and ambiguity, where “the attempt … to gain more information and lucidity and to get evidence finally resulted in the destruction of the detail structure just leaving even more room for obscurity, imagination and speculation.” (Glänzel & Thijs, 2018; cf. Lehmann 2013)

The second, quantitative dimension is a result of quantification and measurement. This implies that a proper balance between feasibility and the demands for the actual assessment of research is needed, where the final granularity choice is a conceptual but quantitatively supported solution.

At ECOOM Leuven, we have found a suitable solution at the granularity level of the 74 subfields (disciplines) according to the Leuven-Budapest classification scheme, which, in turn, is based on the Web of Science Subject Categories. Previous studies by the ECOOM Leuven team (Glänzel & Debackere, 2021; Glänzel et al., 2021) have provided quantitative arguments in support of this choice.

Once the granularity level has been determined, the subject classification – at this level – can be improved. Since subject classification by the providers of the large multidisciplinary citation and abstract databases and their derivatives are usually to a large extent journal based, documents published in multidisciplinary journals need to be individually reassigned to subjects at the chosen level. Such methods have been developed on the basis of the analysis of cited references, for instance, by Glänzel et al. (1999), Glänzel and Schubert (2003) and further elaborated by Milojevič (2020) and Glänzel et al. (2021). Glänzel et al. (2021) have shown that their method allows the individual assignment of about 95% of the documents in question in an automated manner; only the remaining 5% of papers needs to be assigned using other techniques or manually. This assignment strategy is necessary to create the valid and reliable groundwork for the subsequent step, namely the quantification and measurement of the extent of integrated knowledge as represented by scientific documents, and determining the cognitive distance between the individual disciplines as to be discussed in the next section.