Decoding Human Signals: Emotion, Intention, and Argument Detection for Real-World Applications
By Leila MOUDJARI

Leila MOUDJARI will give a seminar on Decoding Human Signals: Emotion, Intention, and Argument Detection for Real-World Applications

Abstract

Understanding human communication goes beyond words—it requires capturing emotions, uncovering intentions, and reasoning over arguments. In this talk, I will present my work at the intersection of these challenges. First, I will discuss emotion detection, showing how models trained on general dialogue datasets adapt to domain-specific contexts such as in-car interactions. Next, I will turn to intention detection, highlighting two use cases: detecting driver intentions in conversational car assistants, and identifying action-oriented intentions in crisis management dialogues. Finally, I will introduce my ongoing work on argument mining for legal reasoning, where we are currently exploring the use of large language models to support the annotation of argumentative structures in judicial texts. Together, these projects illustrate a path toward richer natural language understanding that bridges affect, action, and reasoning, with practical implications across safety-critical, assistive, and legal domains.

Biography

Leila is a researcher and lecturer specializing in artificial intelligence and natural language processing. She holds a PhD in Computer Science, with a focus on machine learning and language technologies, carried out in collaboration with ANITI, CNRS, and IRIT (France). Her research explores applied AI for text, with contributions in subjectivity detection (emotions, hate and hope speech), intention detection in safety-critical and assistive contexts, and the emerging field of argument mining for legal reasoning. She is also dedicated to the development of robust NLP systems for multilingual and low-resource settings, aiming to make language technologies more inclusive and impactful across domains.