Accelerating technology watch by generating an AI-driven taxonomy

Identifying documents belonging to a taxonomy is a very challenging task. In an ideal scenario, this process generally starts with developing a natural language description followed by coding a large number of example documents, which together constitute a robust definition. The process of manually coding documents can be extremely difficult and time consuming. Leveraging AI and various custom tooling, DS has developed an automated workflow to accelerate the process of mapping documents to research area schema, or category based taxonomy. Once completed, this process outputs a scalable machine learning model capable of applying the custom taxonomy to vast document sets quickly. This custom taxonomy can then be made available across publication or awarded grant data to perform custom analytics, and gain insights into existing research through other data enhancements. This work serves as a prime example of how bespoke AI solutions applied to global research intelligence infrastructure can directly support national security and strategic policy objectives in the modern eResearch environment.ABOUT THE AUTHORAnthony Dona is currently employed by Digital Science to inform and direct government organisations across Asia-Pacific on best practices when tracking and measuring the impact of novel research from existing resources, software solutions and databases. Anthony has a specific passion to promote and drive best research practice and optimise novel research outcomes. Previously, Anthony has worked as an analytical chemist/clinical research scientist (at Imperial College London and University of Sydney) specialising in developing and implementing methodologies that marry clinical metadata with comprehensive metabolite measurements from human or animal biofluids. He continues to drive his research interests, and mentor postgraduate students, through an honorary position at the University of Sydney. Furthermore, he is well versed and has interests in developing simple and useful visualisations to interpret complex datasets (in bibliometrics or disease mapping). -----For more information about the eResearch NZ / eRangahau Aotearoa conference, visit: https://eresearchnz.co.nz/