Developing Scalable and Adaptive Methodologies for Measuring Innovation

This dissertation aims to address the challenges associated with traditional measures of innovation by exploring how new methodologies, such as natural language processing, large language models, and emerging data sources, can enhance and complement these conventional metrics. Traditional innovation measures— like patent counts, R&D spending, and product launches may to capture multiple perspective of innovation measures, also such methodologies might struggle to adapt to various context swiftly, and also provide adequate coverage across various sectors regions, or firm type. As a result, these conventional metrics can lead to incomplete assessments, limiting the ability of researchers, policymakers, and industry stakeholders to accurately evaluate and respond to innovation dynamics. In this dissertation, the research objectives focus on two key questions: how can new methodologies, particularly those incorporating big data analytics, be applied a more adaptable and comprehensive measures of innovation and in what ways can these novel methodologies contribute to the field of innovation management studies by providing more actionable metrics? By conducting a thorough literature review and presenting case studies, this research pursues to offer solutions integrating traditional and novel approaches, ultimately delivering additional perspectives for designing innovation strategies and evaluations in various contexts.

This research adopts a critical realist approach and utilizes a pragmatist philosophy, focusing on practical solutions for innovation measurement. The methodology integrates both positivist and interpretivist- constructivist elements to capture objective and subjective realities. The research strategy also employs primarily quantitative methods, based on large-scale sample populations, utilizing various data types and analytics methods across different publications, mainly based on natural language processing. Levering adaptable methodologies, can be tailored to the research question. By conducting multiple case studies, the dissertation demonstrates how novel data sources and analytical techniques, such as web scraping, text mining, and large language models can complement and enhance traditional innovation metrics. The findings shows that more flexible, scalable, and context-oriented indicators enable deeper insights into firm-level innovation, digitalization, paradigm-shifting discoveries, and emerging technological trends.

The theoretical contributions of this dissertation include the creation of novel data-driven indicators for innovation processes and portfolios, enhancing existing metrics, and proposing multidimensional measures for innovation comprehensively. Managerial contributions emphasize the practicality of these novel methodologies for continuous ecosystem monitoring, policymaking, strategic planning, and fostering a culture of data-driven decision making. The dissertation’s findings are contextualized within the existing literature, highlighting its advancements in innovation measurement and proposing future research directions to address current limitations in traditional measures. Future studies could advance text analysis tools, such as advanced information retrieval methods or Knowledge graph to reflect more novel innovation indicators. Also, future research can focus on linking various innovation datasets, at different level, which reveal additional value of innovation data.