Unveiling the knowledge structure of technological forecasting and social change (1969–2020) through an NMF-based hierarchical topic model

Abstract

• The paper presents a longitudinal analysis of the content and topics of the journal technological forecasting and social change. • There are clear bureaus or tranches of literature in the journal corresponding to its long-term mission of technological forecasting and of technology assessment. • The paper demonstrates the use of the non-negative matrix factorization technique on text data. • The paper demonstrates a hierarchical structure of topics is present in the technological forecasting and social change literature, and that this hierarchical structure can be identified through introspection of topic analysis matrices. This article examines the knowledge structure of the journal Technology Forecasting and Social Change (TFSC) from its inception in 1969 until 2020. In this paper we argue that the structure of knowledge in the field of technological forecasting is more complex than a topic model – a bag of words – can in fact effectively reveal. Therefore we propose and demonstrate a hierarchical model that selectively combines topics at varying levels of generality. The resultant analysis, which is based on non-negative matrix factorization, reveals four distinct branches of technology forecasting work, composed of seven distinct topics. Each topic and branch are examined individually through a detailed examination of terms and keywords. Representative works and authors in each of the branches are also identified. The method enables the examination of the complex structure of knowledge in a scientific journal in a succinct representation. The resultant analysis can assist future researchers, enabling them to better position their work, and to better identify the key references across the various subject silos.