Computation of semantic change in scientific concepts: Case study of ``circular economy''

In this chapter we aim to investigate semantic change in a scientific concept underpinned by the evolutionary framework of scientific knowledge production. The aim of this article is threefold. First is to distinguish semantic change computation in scientific concepts from that in core vocabulary and slang. Second is a multi-step analysis combining topic modelling, co-occurence networks and word embeddings, along with a control condition setup thereby presenting a pipeline to compute semantic change in a scientific concept. Third is an analysis of a popular concept in sustainability studies, i.e., ``circular economy'', seeking to advance research on this concept. In order to achieve our objectives, we use topic modelling to detect the point of change in a literature corpus and then we apply two approaches for detecting semantic change: co-occurence networks and word embeddings. Furthermore, we compare the concept with other related concepts in the same semantic field and use word embeddings to detect if the concept has undergone any changes relative to other concepts.