This project investigates how the scientific understanding of the environment has evolved over the last decades. While there has been enormous growth in the scientific understanding of global environmental change, little is known about changes in the environmental sciences themselves: is there a trend towards comprehensive, holistic forms of knowledge or are the sciences becoming increasingly specialized into smaller and smaller niches? Which types of environmental challenges are emerging as dominant in the scientific discourse and which ones are being marginalized? Which linkages and interconnections do researchers construct between them?
The aim of this project is to map changes in the semantic structure of the environmental sciences. We use an original data set consisting of the abstracts of articles published in the top journals in the environmental sciences over the last three decades. Using software for language processing and network analysis, we assess patterns of conceptual co-occurrence, that is: concepts that are being used in the same context. We use concepts as the nodes of our semantic network, with the strength of the connections (“edges”) between nodes being determined by the amount of journal abstracts in which concepts co-occur. Implementing this approach at a large scale allows us to infer changes in semantic structures over time. An initial test run using —10.000 journal abstracts resulted in proof of concept: the method is feasible and can be scaled up to the level of big data. Therefore, the aim for an indicative target of 500.000 journal articles was set.
KTU R&D&I Fund
Period of project implementation: 2018-04-03 - 2018-12-31
Project coordinator: Kaunas University of Technology