Unveiling research landscapes using open research information

Recent studies have shown that OpenAlex, one of the main open databases of scientific publications, has greater coverage than proprietary alternatives both in terms of volume and in terms of diversity. However, it also suffers from some limitations, for example, when it comes to affiliation disambiguation. Our talk will present some of the ways in which we can improve/enrich OpenAlex data through AI. First, we will introduce Affilgood (Duran-Silva et al, 2024), a tool to improve affiliation disambiguation and to enrich its geographical metadata. Second, we will present a semi-automatic methodology to build on-demand classification models based on open data, by leveraging on generative LLMs as well as an open-source tool for collaborative dataset curation. Finally, we will demonstrate how entirely new classifications can be developed and openly shared, an approach that is not possible with closed or proprietary data infrastructures.Furthermore, we will present concrete examples of how collaborative and open innovation approaches are enabling analytical and monitoring capabilities that were previously out of reach. These breakthroughs are driven by access to large-scale open datasets: for example, biomedical analyses using OpenAlex data, which provide extensive coverage across the large biomedical domain.