Nhung T.H. Nguyen

In this project, I focused on developing methods for Named Entity Recognition, Entity Linking and Relation Extraction. The ultimate goal is to build a knowledge graph that contains information about nutrion and health, especially mental health. The resulting graph can then be used for personalised nutrtion.

I was leading research on text mining to automatically detect risk factors and calculate the corresponding risk levels in clinical health records. The main goal of this project is to support underwriters in their risk assement.

This was an EU H2020 project, in which I contributed to the development of an open and interoperable text mining infrastructure. Particularly, we used uimaFIT—a Java framework that allows us to wrap up our text mining tools and make them compatible with tools from other groups. I was involved in two use cases: (1) Extract metabolites and their properties and modes of actions and (2) text mining for curation neuroscience literature.

NCSTOX was funded by cosmetic industry in France to analyse toxicity from text and alleviate their safety assessment process in producing cosmetics. Specifically, we detected molecules, molecule groups and toxicological keywords in several resources, e.g., PubMed, PMC, NCCS Opinions and ToxNet. We then identify relations between plants and molecules, and plants’ organisms and molecules, in order to identify plant extracts that can be used in producing cosmetics. All results have been integrated into the NCSTOX database.

I produced novel tools and resources to curate and analyse biodiversity information for the Philippines to create a Living Atlas: (1) the first terminological repository for biodiversity that was automatically created by applying distributional semantic models to 100 million pages from the Biodiversity Heritage Library, (2) a gold standard for named entities in five categories: species names, habitats, geographical locations, persons and temporal expressions. This data can be used to train tools to extract entities and relations from biodiversity literature.