Drugs4COVID: Combining natural language processing, text mining and knowledge graphs in Health: challenges and a use case

Outline

In this seminar we will describe how we have used a range of state-of-the-art methods, techniques and tools in the areas of Natural Language Processing, text mining and knowledge graphs to build an online system that allows browsing a large corpus of scientific literature that was created and has been maintained since March 2020, with the emergence of the COVID-19 pandemic. After providing a general overview of why and how we built the system, we will go into more depth in areas such as probabilistic topic models and knowledge-graph-based question answering.

Syllabus

• Drugs4COVID: motivation, resources, challenges and steps (including some hands-on activities to browse through the resources and knowledge graphs).
• Probabilistic topic models (including a hands-on activity to create a model from a small corpus).
• Knowledge-graph question answering (including a hands-on activity to understand current opportunities, limitations and challenges).
• ML Demo.cratization initiative (Violette Lepercq)

Assessment Method

Participation during the lecture plus an assignment.

Remarks

This seminar includes an online webinar that will present the HuggingFace initiative through a step-by-step demo about how to build a NLP-powered application.

Recommended Reading

Carlos Badenes-Olmedo, David Chaves-Fraga, María Poveda-Villalón, Ana Iglesias-Molina, Pablo Calleja, Socorro Bernardos, Patricia Martín-Chozas, Alba Fernández-Izquierdo, Elvira Amador-Domínguez, Paola Espinoza-Arias, Luis Pozo, Edna Ruckhaus, Esteban González-Guardia, Raquel Cedazo, Beatriz López-Centeno, Oscar Corcho (2020) Drugs4Covid: Drug-driven Knowledge Exploitation based on Scientific Publications. https://arxiv.org/abs/2012.01953.

Timetable

• 11 March, 15:00-18:00

Lecture Theatre

• A-5001

Tuition Language

English.