Mark Granroth-Wilding is a research associate at the Department of Computer Science, University of Helsinki, with expertise in Natural Language Processing (NLP), Computational Creativity and Computational Music Analysis. His current work concerns unsupervised analysis of low-resourced languages for linguistic typology; topic modelling of noisy, historical data; and Natural Language Generation for news reporting.
He has worked on applying a variety of types of machine learning to different NLP tasks, including extracting information about narrative events from text, and learning representations of symbols in different languages without parallel resources.
Mark has also applied NLP techniques to Computational Creativity, in particular in the What-If Machine project, where he worked on automatic learning of typical event sequences to be used in creative narrative generation. In his PhD, he applied statistical techniques and formal models from NLP to automatic harmonic analysis of music.