Meet Johanna Björklund, a WASP affiliated Associate Professor at Umeå University and project manager of WARA Media & Language. Björklund joined WASP in 2020.
I am the project manager for a WASP Research Arena (WARA): WARA Media & Language. I am also co-PI in a NEST called STING – Synthesis and analysis with Transducers and Invertible Neural Generators, and industrial advisor for a WASP industrial PhD student.
WASP has a vibrant research community, and the program offers some unique funding opportunities.
The program has tremendous agility and the shared network among the WASP faculty is extremely strong.
My background is in theoretical computer science, and this has in combination with recent developments in deep learning taken me into neuro-symbolic reasoning. This means that I am interested in computational models that combine classic, rule-based, reasoning with continuous-state devices, i.e., neural networks. In terms of applications, I focus on the translation of combinations of text, images, and audio into graph-based representations that capture semantic aspects of the input content.
The paper “Improved N-Best Extraction with an Evaluation on Language Data” by Johanna Björklund, Frank Drewes and Anna Jonsson was recently published in the journal Computational Linguistics and will be presented at the 60th Annual Meeting of the Association for Computational Linguistics (ACL 2022).
The hope is that this line of research can combine the best of both worlds of formal reasoning and machine learning, integrating the faculties of deduction and perception. The application to media analysis opens for automation of media workflows and information management.