Sophia Ananiadou is Professor in Computer Science, School of Computer Science, The University of Manchester, Director of the National Centre for Text Mining and a Turing Fellow. Since 2005, she has successfully directed NaCTeM to be a fully sustainable centre, carrying out novel, world-leading research on text mining that then informs the provision of services, tools, resources and infrastructure to a variety of users from translational medicine, biology, biodiversity, humanities, health, and social sciences.
Research she has led has advanced the state of the art in text mining and contributed in novel ways to: automatic extraction of terminology and term variation; development of robust taggers for biomedical text; automatic extraction of events and their interpretation using machine learning methods; development of large scale terminological resources for biomedicine and biodiversity; linking textual evidence with metabolic and signaling pathways; association mining and hypothesis generation; supporting the development of systematic reviews using novel topic modeling and clustering methods and the development of interoperable text mining infrastructure to facilitate all the above applications (Argo). Her team achieved top performance in several NLP and text mining challenges, e.g. BioCreaTive, BioNLP, n2c2, etc. Her h-index is 51 with more than 10,000 citations.
Keynote Speech Title: Enriching Pathway Models Using Text Mining
Summary: Pathway models are valuable resources that help us to understand the various mechanisms underpinning complex biological processes. Their curation is typically carried out through manual inspection of the scientific literature, a knowledge-intensive and laborious task. Text mining methods are used to automate model reconstruction by increasing the speed and reliability of discovery and extracting evidence from the literature. Complex information from the literature is automatically extracted and then mapped to reactions in existing pathway models. Information from the literature (events) can act as corroborative evidence of the validity of these reactions in a model or help to extend it. In addition, by contextualising the textual evidence (extracting uncertainty, negation), we can provide additional confidence measures for linking and ranking information from the literature for model curation and ultimately experimental design. In addition, visual analytics methods can act as the nexus between text mining methods and modellers by providing an interactive way to explore and analyse the statements linked with pathways.