19th Belgium NLP Meetup

Thanks to Sofie Van Landeghem and the folks at Dataroots, we've managed to squeeze in a final Belgium NLP Meetup before the summer holidays. On Monday June 19th, Large Language Models will take center stage: Sofie Van Landeghem (Explosion.ai) will talk about spaCy's LLM extension, Karel D'Oosterlinck (UGent) will show how LLMs can improve drug safety monitoring, and Sophie De Coppel will discuss how Dataroots used LLMs to develop an app that enhances accessibility for disabled individuals. Place to be is the Dataroots office in Leuven, Tiensevest 132, close to the train station. Doors open for drinks at 7pm, talks will start around 7.30pm.

How to integrate large language models into the development and deployment of structured NLP pipelines.
**Sofie Van Landeghem, Explosion.ai**
Due to their impressive natural language capabilities, recent large language models (LLMs) are paving the way for fast prototyping of NLP applications in any business domain. Most practical use-cases however will benefit from a structured, pipeline approach in which LLMs can be complemented with supervised models or even rule-based approaches. In this talk, Sofie showcases how to build such a structured pipeline with the open-source NLP toolbox spaCy, and its recent extension 'spacy-llm'. She'll discuss how to design a production-ready NLP pipeline, while managing different (and occassionally conflicting) performance features such as accuracy, speed, memory usage, reliability, maintainability and customizability.

Improving Drug Safety Monitoring with Natural Language Processing
Karel D'Oosterlinck, UGent
Drug safety monitoring (also called pharmacovigilance) is incredibly important for public safety. Every time we take a medicine, we rely on this process to keep us safe. One important step in the drug safety pipeline is surveying the recent biomedical literature for any evidence of adverse drug events (e.g. a previously unknown side-effect). Today, this is done by a small army of drug safety experts that go through many new biomedical literature each week. Can we improve this process with Natural Language Processing?
In this talk Karel will discuss the result of a collaboration between Ghent University, Stanford University, Aarhus University and Parexel AI labs, aimed at answering this question. Other than outlining the work done, he will aim to give useful tips for anyone trying to get started with an NLP/LM project and outline some interesting directions for future work.

From Words to Answers: Developing a Voice-Powered Intelligent Agent for Querying Data
Sophie De Coppel, Dataroots
Sophie will show how Dataroots developed an an innovative app designed to enhance accessibility for disabled individuals, by using natural language and safeguarded LLMs.