Webinar video https://www.youtube.com/watch?v=j-RTJRCh9oM
Presentation slides
Abstract
While data science initially focused on digital data, it later turned its attention to textual data. Today, advances in automated document processing are among the most significant recent developments in AI, thanks in particular to data mining and machine learning methods. This presentation will illustrate this through the task of stance detection, which aims to identify the position (pro, against, neutral) of a text or author towards a given topic. We will also see that while deep learning, particularly through large language models (LLMs), can very effectively solve this task, it also has serious limitations related to the reproduction or even amplification of biases by the models.
Short Bio
Christine Largeron is a full professor of computer science at Jean Monnet University (France), where she is the head of the RTML team (Responsible and Truthworthly Machine Learning) within the Hubert Curien Laboratory (UMR CNRS 5516). She is also adjunct professor at Edmonton University (Alberta, Canada). Her research focuses on machine learning and data mining on complex data with a particular interest in quality, including fairness and explainability. She is responsible for several research programs, notably the CNRS IEA project “Data mining and Machine Learning for Complex Data Analysis”.
Information at FEP News
More information at News U. Porto