Technological change plays a critical role in industrial- and societal development. As a consequence, modelling, measuring and monitoring the rate and direction of technological change has been extensively studied. However, there is to-date no scalable, cross-domain, and data source agnostic quantitative indicator for technological change that considers a combinatorial process of invention and technology development. This paper develops and empirically tests a network-based method that takes a combinatorial view of technological development as underlying rationale and provides a quantitative indicator for technological change. Unlike prior research, the proposed method allows for the simultaneous inclusion of multiple and diverse types of data sources, i.e. publications, patents, and projects and it uses text-mining analyses based on co-occurrences of terms over time. The novel method proposed here goes beyond reliance on domain experts building custom sets of taxonomies, is applicable across different technological domains, and permits a temporal analysis of changes across years, industries, and countries. This is illustrated using a large database of worldwide bioenergy research and development (R&D) records. Findings include the detection of biofuel generations in the data before they were first mentioned as such in literature. Implications for research, industry and policy are also discussed.