Products & Services based on Semantic Technologies and Natural Language Processing/Understanding for Policy-Making in Science, Technology & Innovation

Science Technology and Innovation (STI) Policy-makers face today technical difficulties that become strategic challenges: data on STI activities is scattered around several non—interoperable sources (e.g., R&D project repositories, patent and bibliometric databases) and it is classified within categories which are i. not transversal across data sources and ii. not relevant for an effective decision—making process in most of the cases. These limitations hinder in turn any benchmarking activity at the institutional and geographical level, preventing a useful monitoring of the effectiveness of the implemented policies.||||Luckily enough, a few transformational trends are opening lines of opportunity to overcome the above limitations: Open government data, the Open science and innovation paradigm and Semantic web technologies are making more (reusable and interoperable) data available in the science and innovation domain. In parallel, Data Science, Artificial Intelligence and Natural Language Processing techniques facilitate a deeper analysis of large amounts of quantitative and text data, allowing for the extraction and development of ad-hoc taxonomies across regions, data sources and traditional classification systems.||||SIRIS Academic is a consultancy firm that provides strategic advice for STI decision making. To do so, we embed technological products in our services that make the most of the above technological trends, in order to aggregate data and analyse it to provide insights and inform strategic decisions.||||Our analytical tools aim at answering to issues such as the identification of STI actors in a given territorial/institutional perimeter, the classification of their research topics in both bottom-up and top down fashion, the discovery of emerging research trends and their alignment with specific societal challenges, as well as benchmarking and matchmaking. To do so, we combine Knowledge Graphs-mediated data access and integration with textual classification techniques based on both Deep Learning and more traditional Natural Language Processing techniques. The solutions we develop offer an array of analytical indicators that help take informed decisions based on granular data, which is typically not available in an aggregated manner, nor categorised with labels which are relevant for political decision makers.||||In this talk, we briefly frame the industry problem we aim at tackling, the solutions we offer to address the market needs and we finally show some practical applications of our methodologies.