In this project the focus is on publications' novelty as an indicator of scientific advancement with regard to collaboration and funding patterns. First, we investigate the relationship between collaborations and novelty in publications. Novelty in scientific publications often incorporates new ideas and opportunities from interdisciplinarity across heterogenous research groups to solve practical problems. Interdisciplinarity, which results from scientific collaboration, is believed to bring out new research outputs, and thus it is reasonable to expect some correlations between collaboration and novelty measures. To the best of our knowledge, however, only few studies have examined the relationship between scientific collaborations and novelty. Specifically, we examine collaborations at the country and various sub-national, i.e. regional, as well as the institutional levels. We measure novelty of articles with keywords' information. Finally, we evaluate the role of funding in research novelty.

The overall goal of this project is to a) to analyse how AI is integrated into the technical knowledge space of regions, and b) to map the creation of scientific knowledge at the global, national and regional scale, and there especially to show the evolution of AI knowledge and again how embedded it is in regional economies. In particular, we seek to build a comprehensive panel of AI patents and scientific publications. Along these lines, we first use identifying keywords to find every AI patent published in Europe between 1985 and 2013. Succinctly, we query the patent abstracts for n-grams like "Genetic Algorithms" or "Neural Networks." We then use this data to map the patents into regions - at the NUTS2 level - and technological codes - the 4-digits CPCs. And next, we borrow methods from network sciences to measure the embeddedness of AI into each region's knowledge space. In doing so, we find that AI "superstar" regions, where AI output is largest, are also those where it's most embedded into the innovation networks. We continue our project to map the development of AI as indicated in scientific publications. We reuse the keywords to identify AI-related publications in the Web of Science. And we employ the metadata within these publications to map the AI evolution across different scales - i.e., the continental, national, and regional levels. Through this exercise, we once again identify potential AI "hotspots" while also distinguishing between places that lead the development of AI from the onset vis-a-vis those that have managed to catch up over time. Besides, we produce networks that illustrate international collaborative efforts in AI knowledge creation via co-authorship across nations and via the evolution of keywords' co-occurrence across three decades. It is evident how these networks become denser with time and how they differ across regions. Therefore, the project aims to enhance our understanding of where and how AI knowledge emerges. Although the results are still exploratory, the project is the start of a reliable database to answer the big questions about AI systems. In particular, we hope to encourage and assist future research efforts addressing the creation of AI and its consequences.  The project relates to both the ERC TechEvo and SFI SciTechSpace funded research grants.